Fix include paths and compiling in Linux. Externals soundtouch is 1.7.1, while Ubuntu 12.10 is 1.6.x. Externals soundtouch is compiled with integer samples, while ubuntu is compiled with float samples. Float samples is probably the more common route. If you're going to use soundtouch, you should probably use SAMPLETYPE instead of explicitly choosing short. This probably breaks the windows build since its includes aren't setup.

Externals/soundtouch/AAFilter.cpp

@@ -0,0 +1,184 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// FIR low-pass (anti-alias) filter with filter coefficient design routine and
+/// MMX optimization. 
+/// 
+/// Anti-alias filter is used to prevent folding of high frequencies when 
+/// transposing the sample rate with interpolation.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2009-01-11 13:34:24 +0200 (Sun, 11 Jan 2009) $
+// File revision : $Revision: 4 $
+//
+// $Id: AAFilter.cpp 45 2009-01-11 11:34:24Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <memory.h>
+#include <assert.h>
+#include <math.h>
+#include <stdlib.h>
+#include "AAFilter.h"
+#include "FIRFilter.h"
+
+using namespace soundtouch;
+
+#define PI        3.141592655357989
+#define TWOPI    (2 * PI)
+
+/*****************************************************************************
+ *
+ * Implementation of the class 'AAFilter'
+ *
+ *****************************************************************************/
+
+AAFilter::AAFilter(uint len)
+{
+    pFIR = FIRFilter::newInstance();
+    cutoffFreq = 0.5;
+    setLength(len);
+}
+
+
+
+AAFilter::~AAFilter()
+{
+    delete pFIR;
+}
+
+
+
+// Sets new anti-alias filter cut-off edge frequency, scaled to
+// sampling frequency (nyquist frequency = 0.5).
+// The filter will cut frequencies higher than the given frequency.
+void AAFilter::setCutoffFreq(double newCutoffFreq)
+{
+    cutoffFreq = newCutoffFreq;
+    calculateCoeffs();
+}
+
+
+
+// Sets number of FIR filter taps
+void AAFilter::setLength(uint newLength)
+{
+    length = newLength;
+    calculateCoeffs();
+}
+
+
+
+// Calculates coefficients for a low-pass FIR filter using Hamming window
+void AAFilter::calculateCoeffs()
+{
+    uint i;
+    double cntTemp, temp, tempCoeff,h, w;
+    double fc2, wc;
+    double scaleCoeff, sum;
+    double *work;
+    SAMPLETYPE *coeffs;
+
+    assert(length >= 2);
+    assert(length % 4 == 0);
+    assert(cutoffFreq >= 0);
+    assert(cutoffFreq <= 0.5);
+
+    work = new double[length];
+    coeffs = new SAMPLETYPE[length];
+
+    fc2 = 2.0 * cutoffFreq; 
+    wc = PI * fc2;
+    tempCoeff = TWOPI / (double)length;
+
+    sum = 0;
+    for (i = 0; i < length; i ++) 
+    {
+        cntTemp = (double)i - (double)(length / 2);
+
+        temp = cntTemp * wc;
+        if (temp != 0) 
+        {
+            h = fc2 * sin(temp) / temp;                     // sinc function
+        } 
+        else 
+        {
+            h = 1.0;
+        }
+        w = 0.54 + 0.46 * cos(tempCoeff * cntTemp);       // hamming window
+
+        temp = w * h;
+        work[i] = temp;
+
+        // calc net sum of coefficients 
+        sum += temp;
+    }
+
+    // ensure the sum of coefficients is larger than zero
+    assert(sum > 0);
+
+    // ensure we've really designed a lowpass filter...
+    assert(work[length/2] > 0);
+    assert(work[length/2 + 1] > -1e-6);
+    assert(work[length/2 - 1] > -1e-6);
+
+    // Calculate a scaling coefficient in such a way that the result can be
+    // divided by 16384
+    scaleCoeff = 16384.0f / sum;
+
+    for (i = 0; i < length; i ++) 
+    {
+        // scale & round to nearest integer
+        temp = work[i] * scaleCoeff;
+        temp += (temp >= 0) ? 0.5 : -0.5;
+        // ensure no overfloods
+        assert(temp >= -32768 && temp <= 32767);
+        coeffs[i] = (SAMPLETYPE)temp;
+    }
+
+    // Set coefficients. Use divide factor 14 => divide result by 2^14 = 16384
+    pFIR->setCoefficients(coeffs, length, 14);
+
+    delete[] work;
+    delete[] coeffs;
+}
+
+
+// Applies the filter to the given sequence of samples. 
+// Note : The amount of outputted samples is by value of 'filter length' 
+// smaller than the amount of input samples.
+uint AAFilter::evaluate(SAMPLETYPE *dest, const SAMPLETYPE *src, uint numSamples, uint numChannels) const
+{
+    return pFIR->evaluate(dest, src, numSamples, numChannels);
+}
+
+
+uint AAFilter::getLength() const
+{
+    return pFIR->getLength();
+}

Externals/soundtouch/AAFilter.h

@@ -0,0 +1,91 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Sampled sound tempo changer/time stretch algorithm. Changes the sound tempo 
+/// while maintaining the original pitch by using a time domain WSOLA-like method 
+/// with several performance-increasing tweaks.
+///
+/// Anti-alias filter is used to prevent folding of high frequencies when 
+/// transposing the sample rate with interpolation.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2008-02-10 18:26:55 +0200 (Sun, 10 Feb 2008) $
+// File revision : $Revision: 4 $
+//
+// $Id: AAFilter.h 11 2008-02-10 16:26:55Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef AAFilter_H
+#define AAFilter_H
+
+#include "STTypes.h"
+
+namespace soundtouch
+{
+
+class AAFilter
+{
+protected:
+    class FIRFilter *pFIR;
+
+    /// Low-pass filter cut-off frequency, negative = invalid
+    double cutoffFreq;
+
+    /// num of filter taps
+    uint length;
+
+    /// Calculate the FIR coefficients realizing the given cutoff-frequency
+    void calculateCoeffs();
+public:
+    AAFilter(uint length);
+
+    ~AAFilter();
+
+    /// Sets new anti-alias filter cut-off edge frequency, scaled to sampling 
+    /// frequency (nyquist frequency = 0.5). The filter will cut off the 
+    /// frequencies than that.
+    void setCutoffFreq(double newCutoffFreq);
+
+    /// Sets number of FIR filter taps, i.e. ~filter complexity
+    void setLength(uint newLength);
+
+    uint getLength() const;
+
+    /// Applies the filter to the given sequence of samples. 
+    /// Note : The amount of outputted samples is by value of 'filter length' 
+    /// smaller than the amount of input samples.
+    uint evaluate(SAMPLETYPE *dest, 
+                  const SAMPLETYPE *src, 
+                  uint numSamples, 
+                  uint numChannels) const;
+};
+
+}
+
+#endif

Externals/soundtouch/BPMDetect.cpp

@@ -0,0 +1,370 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Beats-per-minute (BPM) detection routine.
+///
+/// The beat detection algorithm works as follows:
+/// - Use function 'inputSamples' to input a chunks of samples to the class for
+///   analysis. It's a good idea to enter a large sound file or stream in smallish
+///   chunks of around few kilosamples in order not to extinguish too much RAM memory.
+/// - Inputted sound data is decimated to approx 500 Hz to reduce calculation burden,
+///   which is basically ok as low (bass) frequencies mostly determine the beat rate.
+///   Simple averaging is used for anti-alias filtering because the resulting signal
+///   quality isn't of that high importance.
+/// - Decimated sound data is enveloped, i.e. the amplitude shape is detected by
+///   taking absolute value that's smoothed by sliding average. Signal levels that
+///   are below a couple of times the general RMS amplitude level are cut away to
+///   leave only notable peaks there.
+/// - Repeating sound patterns (e.g. beats) are detected by calculating short-term 
+///   autocorrelation function of the enveloped signal.
+/// - After whole sound data file has been analyzed as above, the bpm level is 
+///   detected by function 'getBpm' that finds the highest peak of the autocorrelation 
+///   function, calculates it's precise location and converts this reading to bpm's.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-08-30 22:45:25 +0300 (Thu, 30 Aug 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: BPMDetect.cpp 149 2012-08-30 19:45:25Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <math.h>
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+#include "FIFOSampleBuffer.h"
+#include "PeakFinder.h"
+#include "BPMDetect.h"
+
+using namespace soundtouch;
+
+#define INPUT_BLOCK_SAMPLES       2048
+#define DECIMATED_BLOCK_SAMPLES   256
+
+/// decay constant for calculating RMS volume sliding average approximation 
+/// (time constant is about 10 sec)
+const float avgdecay = 0.99986f;
+
+/// Normalization coefficient for calculating RMS sliding average approximation.
+const float avgnorm = (1 - avgdecay);
+
+
+////////////////////////////////////////////////////////////////////////////////
+
+// Enable following define to create bpm analysis file:
+
+// #define _CREATE_BPM_DEBUG_FILE
+
+#ifdef _CREATE_BPM_DEBUG_FILE
+
+    #define DEBUGFILE_NAME  "c:\\temp\\soundtouch-bpm-debug.txt"
+
+    static void _SaveDebugData(const float *data, int minpos, int maxpos, double coeff)
+    {
+        FILE *fptr = fopen(DEBUGFILE_NAME, "wt");
+        int i;
+
+        if (fptr)
+        {
+            printf("\n\nWriting BPM debug data into file " DEBUGFILE_NAME "\n\n");
+            for (i = minpos; i < maxpos; i ++)
+            {
+                fprintf(fptr, "%d\t%.1lf\t%f\n", i, coeff / (double)i, data[i]);
+            }
+            fclose(fptr);
+        }
+    }
+#else
+    #define _SaveDebugData(a,b,c,d)
+#endif
+
+////////////////////////////////////////////////////////////////////////////////
+
+
+BPMDetect::BPMDetect(int numChannels, int aSampleRate)
+{
+    this->sampleRate = aSampleRate;
+    this->channels = numChannels;
+
+    decimateSum = 0;
+    decimateCount = 0;
+
+    envelopeAccu = 0;
+
+    // Initialize RMS volume accumulator to RMS level of 1500 (out of 32768) that's
+    // safe initial RMS signal level value for song data. This value is then adapted
+    // to the actual level during processing.
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+    // integer samples
+    RMSVolumeAccu = (1500 * 1500) / avgnorm;
+#else
+    // float samples, scaled to range [-1..+1[
+    RMSVolumeAccu = (0.045f * 0.045f) / avgnorm;
+#endif
+
+    // choose decimation factor so that result is approx. 1000 Hz
+    decimateBy = sampleRate / 1000;
+    assert(decimateBy > 0);
+    assert(INPUT_BLOCK_SAMPLES < decimateBy * DECIMATED_BLOCK_SAMPLES);
+
+    // Calculate window length & starting item according to desired min & max bpms
+    windowLen = (60 * sampleRate) / (decimateBy * MIN_BPM);
+    windowStart = (60 * sampleRate) / (decimateBy * MAX_BPM);
+
+    assert(windowLen > windowStart);
+
+    // allocate new working objects
+    xcorr = new float[windowLen];
+    memset(xcorr, 0, windowLen * sizeof(float));
+
+    // allocate processing buffer
+    buffer = new FIFOSampleBuffer();
+    // we do processing in mono mode
+    buffer->setChannels(1);
+    buffer->clear();
+}
+
+
+
+BPMDetect::~BPMDetect()
+{
+    delete[] xcorr;
+    delete buffer;
+}
+
+
+
+/// convert to mono, low-pass filter & decimate to about 500 Hz. 
+/// return number of outputted samples.
+///
+/// Decimation is used to remove the unnecessary frequencies and thus to reduce 
+/// the amount of data needed to be processed as calculating autocorrelation 
+/// function is a very-very heavy operation.
+///
+/// Anti-alias filtering is done simply by averaging the samples. This is really a 
+/// poor-man's anti-alias filtering, but it's not so critical in this kind of application
+/// (it'd also be difficult to design a high-quality filter with steep cut-off at very 
+/// narrow band)
+int BPMDetect::decimate(SAMPLETYPE *dest, const SAMPLETYPE *src, int numsamples)
+{
+    int count, outcount;
+    LONG_SAMPLETYPE out;
+
+    assert(channels > 0);
+    assert(decimateBy > 0);
+    outcount = 0;
+    for (count = 0; count < numsamples; count ++) 
+    {
+        int j;
+
+        // convert to mono and accumulate
+        for (j = 0; j < channels; j ++)
+        {
+            decimateSum += src[j];
+        }
+        src += j;
+
+        decimateCount ++;
+        if (decimateCount >= decimateBy) 
+        {
+            // Store every Nth sample only
+            out = (LONG_SAMPLETYPE)(decimateSum / (decimateBy * channels));
+            decimateSum = 0;
+            decimateCount = 0;
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+            // check ranges for sure (shouldn't actually be necessary)
+            if (out > 32767) 
+            {
+                out = 32767;
+            } 
+            else if (out < -32768) 
+            {
+                out = -32768;
+            }
+#endif // SOUNDTOUCH_INTEGER_SAMPLES
+            dest[outcount] = (SAMPLETYPE)out;
+            outcount ++;
+        }
+    }
+    return outcount;
+}
+
+
+
+// Calculates autocorrelation function of the sample history buffer
+void BPMDetect::updateXCorr(int process_samples)
+{
+    int offs;
+    SAMPLETYPE *pBuffer;
+    
+    assert(buffer->numSamples() >= (uint)(process_samples + windowLen));
+
+    pBuffer = buffer->ptrBegin();
+    for (offs = windowStart; offs < windowLen; offs ++) 
+    {
+        LONG_SAMPLETYPE sum;
+        int i;
+
+        sum = 0;
+        for (i = 0; i < process_samples; i ++) 
+        {
+            sum += pBuffer[i] * pBuffer[i + offs];    // scaling the sub-result shouldn't be necessary
+        }
+//        xcorr[offs] *= xcorr_decay;   // decay 'xcorr' here with suitable coefficients 
+                                        // if it's desired that the system adapts automatically to
+                                        // various bpms, e.g. in processing continouos music stream.
+                                        // The 'xcorr_decay' should be a value that's smaller than but 
+                                        // close to one, and should also depend on 'process_samples' value.
+
+        xcorr[offs] += (float)sum;
+    }
+}
+
+
+// Calculates envelope of the sample data
+void BPMDetect::calcEnvelope(SAMPLETYPE *samples, int numsamples) 
+{
+    const static double decay = 0.7f;               // decay constant for smoothing the envelope
+    const static double norm = (1 - decay);
+
+    int i;
+    LONG_SAMPLETYPE out;
+    double val;
+
+    for (i = 0; i < numsamples; i ++) 
+    {
+        // calc average RMS volume
+        RMSVolumeAccu *= avgdecay;
+        val = (float)fabs((float)samples[i]);
+        RMSVolumeAccu += val * val;
+
+        // cut amplitudes that are below cutoff ~2 times RMS volume
+        // (we're interested in peak values, not the silent moments)
+        if (val < 0.5 * sqrt(RMSVolumeAccu * avgnorm))
+        {
+            val = 0;
+        }
+
+        // smooth amplitude envelope
+        envelopeAccu *= decay;
+        envelopeAccu += val;
+        out = (LONG_SAMPLETYPE)(envelopeAccu * norm);
+
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+        // cut peaks (shouldn't be necessary though)
+        if (out > 32767) out = 32767;
+#endif // SOUNDTOUCH_INTEGER_SAMPLES
+        samples[i] = (SAMPLETYPE)out;
+    }
+}
+
+
+
+void BPMDetect::inputSamples(const SAMPLETYPE *samples, int numSamples)
+{
+    SAMPLETYPE decimated[DECIMATED_BLOCK_SAMPLES];
+
+    // iterate so that max INPUT_BLOCK_SAMPLES processed per iteration
+    while (numSamples > 0)
+    {
+        int block;
+        int decSamples;
+
+        block = (numSamples > INPUT_BLOCK_SAMPLES) ? INPUT_BLOCK_SAMPLES : numSamples;
+
+        // decimate. note that converts to mono at the same time
+        decSamples = decimate(decimated, samples, block);
+        samples += block * channels;
+        numSamples -= block;
+
+        // envelope new samples and add them to buffer
+        calcEnvelope(decimated, decSamples);
+        buffer->putSamples(decimated, decSamples);
+    }
+
+    // when the buffer has enought samples for processing...
+    if ((int)buffer->numSamples() > windowLen) 
+    {
+        int processLength;
+
+        // how many samples are processed
+        processLength = (int)buffer->numSamples() - windowLen;
+
+        // ... calculate autocorrelations for oldest samples...
+        updateXCorr(processLength);
+        // ... and remove them from the buffer
+        buffer->receiveSamples(processLength);
+    }
+}
+
+
+
+void BPMDetect::removeBias()
+{
+    int i;
+    float minval = 1e12f;   // arbitrary large number
+
+    for (i = windowStart; i < windowLen; i ++)
+    {
+        if (xcorr[i] < minval)
+        {
+            minval = xcorr[i];
+        }
+    }
+
+    for (i = windowStart; i < windowLen; i ++)
+    {
+        xcorr[i] -= minval;
+    }
+}
+
+
+float BPMDetect::getBpm()
+{
+    double peakPos;
+    double coeff;
+    PeakFinder peakFinder;
+
+    coeff = 60.0 * ((double)sampleRate / (double)decimateBy);
+
+    // save bpm debug analysis data if debug data enabled
+    _SaveDebugData(xcorr, windowStart, windowLen, coeff);
+
+    // remove bias from xcorr data
+    removeBias();
+
+    // find peak position
+    peakPos = peakFinder.detectPeak(xcorr, windowStart, windowLen);
+
+    assert(decimateBy != 0);
+    if (peakPos < 1e-9) return 0.0; // detection failed.
+
+    // calculate BPM
+    return (float) (coeff / peakPos);
+}

Externals/soundtouch/BPMDetect.h

@@ -0,0 +1,164 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Beats-per-minute (BPM) detection routine.
+///
+/// The beat detection algorithm works as follows:
+/// - Use function 'inputSamples' to input a chunks of samples to the class for
+///   analysis. It's a good idea to enter a large sound file or stream in smallish
+///   chunks of around few kilosamples in order not to extinguish too much RAM memory.
+/// - Input sound data is decimated to approx 500 Hz to reduce calculation burden,
+///   which is basically ok as low (bass) frequencies mostly determine the beat rate.
+///   Simple averaging is used for anti-alias filtering because the resulting signal
+///   quality isn't of that high importance.
+/// - Decimated sound data is enveloped, i.e. the amplitude shape is detected by
+///   taking absolute value that's smoothed by sliding average. Signal levels that
+///   are below a couple of times the general RMS amplitude level are cut away to
+///   leave only notable peaks there.
+/// - Repeating sound patterns (e.g. beats) are detected by calculating short-term 
+///   autocorrelation function of the enveloped signal.
+/// - After whole sound data file has been analyzed as above, the bpm level is 
+///   detected by function 'getBpm' that finds the highest peak of the autocorrelation 
+///   function, calculates it's precise location and converts this reading to bpm's.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-08-30 22:53:44 +0300 (Thu, 30 Aug 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: BPMDetect.h 150 2012-08-30 19:53:44Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef _BPMDetect_H_
+#define _BPMDetect_H_
+
+#include "STTypes.h"
+#include "FIFOSampleBuffer.h"
+
+namespace soundtouch
+{
+
+/// Minimum allowed BPM rate. Used to restrict accepted result above a reasonable limit.
+#define MIN_BPM 29
+
+/// Maximum allowed BPM rate. Used to restrict accepted result below a reasonable limit.
+#define MAX_BPM 200
+
+
+/// Class for calculating BPM rate for audio data.
+class BPMDetect
+{
+protected:
+    /// Auto-correlation accumulator bins.
+    float *xcorr;
+    
+    /// Amplitude envelope sliding average approximation level accumulator
+    double envelopeAccu;
+
+    /// RMS volume sliding average approximation level accumulator
+    double RMSVolumeAccu;
+
+    /// Sample average counter.
+    int decimateCount;
+
+    /// Sample average accumulator for FIFO-like decimation.
+    soundtouch::LONG_SAMPLETYPE decimateSum;
+
+    /// Decimate sound by this coefficient to reach approx. 500 Hz.
+    int decimateBy;
+
+    /// Auto-correlation window length
+    int windowLen;
+
+    /// Number of channels (1 = mono, 2 = stereo)
+    int channels;
+
+    /// sample rate
+    int sampleRate;
+
+    /// Beginning of auto-correlation window: Autocorrelation isn't being updated for
+    /// the first these many correlation bins.
+    int windowStart;
+ 
+    /// FIFO-buffer for decimated processing samples.
+    soundtouch::FIFOSampleBuffer *buffer;
+
+    /// Updates auto-correlation function for given number of decimated samples that 
+    /// are read from the internal 'buffer' pipe (samples aren't removed from the pipe 
+    /// though).
+    void updateXCorr(int process_samples      /// How many samples are processed.
+                     );
+
+    /// Decimates samples to approx. 500 Hz.
+    ///
+    /// \return Number of output samples.
+    int decimate(soundtouch::SAMPLETYPE *dest,      ///< Destination buffer
+                 const soundtouch::SAMPLETYPE *src, ///< Source sample buffer
+                 int numsamples                     ///< Number of source samples.
+                 );
+
+    /// Calculates amplitude envelope for the buffer of samples.
+    /// Result is output to 'samples'.
+    void calcEnvelope(soundtouch::SAMPLETYPE *samples,  ///< Pointer to input/output data buffer
+                      int numsamples                    ///< Number of samples in buffer
+                      );
+
+    /// remove constant bias from xcorr data
+    void removeBias();
+
+public:
+    /// Constructor.
+    BPMDetect(int numChannels,  ///< Number of channels in sample data.
+              int sampleRate    ///< Sample rate in Hz.
+              );
+
+    /// Destructor.
+    virtual ~BPMDetect();
+
+    /// Inputs a block of samples for analyzing: Envelopes the samples and then
+    /// updates the autocorrelation estimation. When whole song data has been input
+    /// in smaller blocks using this function, read the resulting bpm with 'getBpm' 
+    /// function. 
+    /// 
+    /// Notice that data in 'samples' array can be disrupted in processing.
+    void inputSamples(const soundtouch::SAMPLETYPE *samples,    ///< Pointer to input/working data buffer
+                      int numSamples                            ///< Number of samples in buffer
+                      );
+
+
+    /// Analyzes the results and returns the BPM rate. Use this function to read result
+    /// after whole song data has been input to the class by consecutive calls of
+    /// 'inputSamples' function.
+    ///
+    /// \return Beats-per-minute rate, or zero if detection failed.
+    float getBpm();
+};
+
+}
+
+#endif // _BPMDetect_H_

Externals/soundtouch/CMakeLists.txt

@@ -0,0 +1,15 @@
+set(SRCS
+	AAFilter.cpp
+	BPMDetect.cpp
+	cpu_detect_x86.cpp
+	FIFOSampleBuffer.cpp
+	FIRFilter.cpp
+	mmx_optimized.cpp
+	PeakFinder.cpp
+	RateTransposer.cpp
+	SoundTouch.cpp
+	sse_optimized.cpp
+	TDStretch.cpp
+	)
+
+add_library(SoundTouch STATIC ${SRCS})

Externals/soundtouch/FIFOSampleBuffer.cpp

@@ -0,0 +1,274 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// A buffer class for temporarily storaging sound samples, operates as a 
+/// first-in-first-out pipe.
+///
+/// Samples are added to the end of the sample buffer with the 'putSamples' 
+/// function, and are received from the beginning of the buffer by calling
+/// the 'receiveSamples' function. The class automatically removes the 
+/// outputted samples from the buffer, as well as grows the buffer size 
+/// whenever necessary.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-11-08 20:53:01 +0200 (Thu, 08 Nov 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: FIFOSampleBuffer.cpp 160 2012-11-08 18:53:01Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <stdlib.h>
+#include <memory.h>
+#include <string.h>
+#include <assert.h>
+
+#include "FIFOSampleBuffer.h"
+
+using namespace soundtouch;
+
+// Constructor
+FIFOSampleBuffer::FIFOSampleBuffer(int numChannels)
+{
+    assert(numChannels > 0);
+    sizeInBytes = 0; // reasonable initial value
+    buffer = NULL;
+    bufferUnaligned = NULL;
+    samplesInBuffer = 0;
+    bufferPos = 0;
+    channels = (uint)numChannels;
+    ensureCapacity(32);     // allocate initial capacity 
+}
+
+
+// destructor
+FIFOSampleBuffer::~FIFOSampleBuffer()
+{
+    delete[] bufferUnaligned;
+    bufferUnaligned = NULL;
+    buffer = NULL;
+}
+
+
+// Sets number of channels, 1 = mono, 2 = stereo
+void FIFOSampleBuffer::setChannels(int numChannels)
+{
+    uint usedBytes;
+
+    assert(numChannels > 0);
+    usedBytes = channels * samplesInBuffer;
+    channels = (uint)numChannels;
+    samplesInBuffer = usedBytes / channels;
+}
+
+
+// if output location pointer 'bufferPos' isn't zero, 'rewinds' the buffer and
+// zeroes this pointer by copying samples from the 'bufferPos' pointer 
+// location on to the beginning of the buffer.
+void FIFOSampleBuffer::rewind()
+{
+    if (buffer && bufferPos) 
+    {
+        memmove(buffer, ptrBegin(), sizeof(SAMPLETYPE) * channels * samplesInBuffer);
+        bufferPos = 0;
+    }
+}
+
+
+// Adds 'numSamples' pcs of samples from the 'samples' memory position to 
+// the sample buffer.
+void FIFOSampleBuffer::putSamples(const SAMPLETYPE *samples, uint nSamples)
+{
+    memcpy(ptrEnd(nSamples), samples, sizeof(SAMPLETYPE) * nSamples * channels);
+    samplesInBuffer += nSamples;
+}
+
+
+// Increases the number of samples in the buffer without copying any actual
+// samples.
+//
+// This function is used to update the number of samples in the sample buffer
+// when accessing the buffer directly with 'ptrEnd' function. Please be 
+// careful though!
+void FIFOSampleBuffer::putSamples(uint nSamples)
+{
+    uint req;
+
+    req = samplesInBuffer + nSamples;
+    ensureCapacity(req);
+    samplesInBuffer += nSamples;
+}
+
+
+// Returns a pointer to the end of the used part of the sample buffer (i.e. 
+// where the new samples are to be inserted). This function may be used for 
+// inserting new samples into the sample buffer directly. Please be careful! 
+//
+// Parameter 'slackCapacity' tells the function how much free capacity (in
+// terms of samples) there _at least_ should be, in order to the caller to
+// succesfully insert all the required samples to the buffer. When necessary, 
+// the function grows the buffer size to comply with this requirement.
+//
+// When using this function as means for inserting new samples, also remember 
+// to increase the sample count afterwards, by calling  the 
+// 'putSamples(numSamples)' function.
+SAMPLETYPE *FIFOSampleBuffer::ptrEnd(uint slackCapacity) 
+{
+    ensureCapacity(samplesInBuffer + slackCapacity);
+    return buffer + samplesInBuffer * channels;
+}
+
+
+// Returns a pointer to the beginning of the currently non-outputted samples. 
+// This function is provided for accessing the output samples directly. 
+// Please be careful!
+//
+// When using this function to output samples, also remember to 'remove' the
+// outputted samples from the buffer by calling the 
+// 'receiveSamples(numSamples)' function
+SAMPLETYPE *FIFOSampleBuffer::ptrBegin()
+{
+    assert(buffer);
+    return buffer + bufferPos * channels;
+}
+
+
+// Ensures that the buffer has enought capacity, i.e. space for _at least_
+// 'capacityRequirement' number of samples. The buffer is grown in steps of
+// 4 kilobytes to eliminate the need for frequently growing up the buffer,
+// as well as to round the buffer size up to the virtual memory page size.
+void FIFOSampleBuffer::ensureCapacity(uint capacityRequirement)
+{
+    SAMPLETYPE *tempUnaligned, *temp;
+
+    if (capacityRequirement > getCapacity()) 
+    {
+        // enlarge the buffer in 4kbyte steps (round up to next 4k boundary)
+        sizeInBytes = (capacityRequirement * channels * sizeof(SAMPLETYPE) + 4095) & (uint)-4096;
+        assert(sizeInBytes % 2 == 0);
+        tempUnaligned = new SAMPLETYPE[sizeInBytes / sizeof(SAMPLETYPE) + 16 / sizeof(SAMPLETYPE)];
+        if (tempUnaligned == NULL)
+        {
+            ST_THROW_RT_ERROR("Couldn't allocate memory!\n");
+        }
+        // Align the buffer to begin at 16byte cache line boundary for optimal performance
+        temp = (SAMPLETYPE *)SOUNDTOUCH_ALIGN_POINTER_16(tempUnaligned);
+        if (samplesInBuffer)
+        {
+            memcpy(temp, ptrBegin(), samplesInBuffer * channels * sizeof(SAMPLETYPE));
+        }
+        delete[] bufferUnaligned;
+        buffer = temp;
+        bufferUnaligned = tempUnaligned;
+        bufferPos = 0;
+    } 
+    else 
+    {
+        // simply rewind the buffer (if necessary)
+        rewind();
+    }
+}
+
+
+// Returns the current buffer capacity in terms of samples
+uint FIFOSampleBuffer::getCapacity() const
+{
+    return sizeInBytes / (channels * sizeof(SAMPLETYPE));
+}
+
+
+// Returns the number of samples currently in the buffer
+uint FIFOSampleBuffer::numSamples() const
+{
+    return samplesInBuffer;
+}
+
+
+// Output samples from beginning of the sample buffer. Copies demanded number
+// of samples to output and removes them from the sample buffer. If there
+// are less than 'numsample' samples in the buffer, returns all available.
+//
+// Returns number of samples copied.
+uint FIFOSampleBuffer::receiveSamples(SAMPLETYPE *output, uint maxSamples)
+{
+    uint num;
+
+    num = (maxSamples > samplesInBuffer) ? samplesInBuffer : maxSamples;
+
+	memcpy(output, ptrBegin(), channels * sizeof(SAMPLETYPE) * num);
+    return receiveSamples(num);
+}
+
+
+// Removes samples from the beginning of the sample buffer without copying them
+// anywhere. Used to reduce the number of samples in the buffer, when accessing
+// the sample buffer with the 'ptrBegin' function.
+uint FIFOSampleBuffer::receiveSamples(uint maxSamples)
+{
+    if (maxSamples >= samplesInBuffer)
+    {
+        uint temp;
+
+        temp = samplesInBuffer;
+        samplesInBuffer = 0;
+        return temp;
+    }
+
+    samplesInBuffer -= maxSamples;
+    bufferPos += maxSamples;
+
+    return maxSamples;
+}
+
+
+// Returns nonzero if the sample buffer is empty
+int FIFOSampleBuffer::isEmpty() const
+{
+    return (samplesInBuffer == 0) ? 1 : 0;
+}
+
+
+// Clears the sample buffer
+void FIFOSampleBuffer::clear()
+{
+    samplesInBuffer = 0;
+    bufferPos = 0;
+}
+
+
+/// allow trimming (downwards) amount of samples in pipeline.
+/// Returns adjusted amount of samples
+uint FIFOSampleBuffer::adjustAmountOfSamples(uint numSamples)
+{
+    if (numSamples < samplesInBuffer)
+    {
+        samplesInBuffer = numSamples;
+    }
+    return samplesInBuffer;
+}
+

Externals/soundtouch/FIFOSampleBuffer.h

@@ -0,0 +1,178 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// A buffer class for temporarily storaging sound samples, operates as a 
+/// first-in-first-out pipe.
+///
+/// Samples are added to the end of the sample buffer with the 'putSamples' 
+/// function, and are received from the beginning of the buffer by calling
+/// the 'receiveSamples' function. The class automatically removes the 
+/// output samples from the buffer as well as grows the storage size 
+/// whenever necessary.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-06-13 22:29:53 +0300 (Wed, 13 Jun 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: FIFOSampleBuffer.h 143 2012-06-13 19:29:53Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef FIFOSampleBuffer_H
+#define FIFOSampleBuffer_H
+
+#include "FIFOSamplePipe.h"
+
+namespace soundtouch
+{
+
+/// Sample buffer working in FIFO (first-in-first-out) principle. The class takes
+/// care of storage size adjustment and data moving during input/output operations.
+///
+/// Notice that in case of stereo audio, one sample is considered to consist of 
+/// both channel data.
+class FIFOSampleBuffer : public FIFOSamplePipe
+{
+private:
+    /// Sample buffer.
+    SAMPLETYPE *buffer;
+
+    // Raw unaligned buffer memory. 'buffer' is made aligned by pointing it to first
+    // 16-byte aligned location of this buffer
+    SAMPLETYPE *bufferUnaligned;
+
+    /// Sample buffer size in bytes
+    uint sizeInBytes;
+
+    /// How many samples are currently in buffer.
+    uint samplesInBuffer;
+
+    /// Channels, 1=mono, 2=stereo.
+    uint channels;
+
+    /// Current position pointer to the buffer. This pointer is increased when samples are 
+    /// removed from the pipe so that it's necessary to actually rewind buffer (move data)
+    /// only new data when is put to the pipe.
+    uint bufferPos;
+
+    /// Rewind the buffer by moving data from position pointed by 'bufferPos' to real 
+    /// beginning of the buffer.
+    void rewind();
+
+    /// Ensures that the buffer has capacity for at least this many samples.
+    void ensureCapacity(uint capacityRequirement);
+
+    /// Returns current capacity.
+    uint getCapacity() const;
+
+public:
+
+    /// Constructor
+    FIFOSampleBuffer(int numChannels = 2     ///< Number of channels, 1=mono, 2=stereo.
+                                              ///< Default is stereo.
+                     );
+
+    /// destructor
+    ~FIFOSampleBuffer();
+
+    /// Returns a pointer to the beginning of the output samples. 
+    /// This function is provided for accessing the output samples directly. 
+    /// Please be careful for not to corrupt the book-keeping!
+    ///
+    /// When using this function to output samples, also remember to 'remove' the
+    /// output samples from the buffer by calling the 
+    /// 'receiveSamples(numSamples)' function
+    virtual SAMPLETYPE *ptrBegin();
+
+    /// Returns a pointer to the end of the used part of the sample buffer (i.e. 
+    /// where the new samples are to be inserted). This function may be used for 
+    /// inserting new samples into the sample buffer directly. Please be careful
+    /// not corrupt the book-keeping!
+    ///
+    /// When using this function as means for inserting new samples, also remember 
+    /// to increase the sample count afterwards, by calling  the 
+    /// 'putSamples(numSamples)' function.
+    SAMPLETYPE *ptrEnd(
+                uint slackCapacity   ///< How much free capacity (in samples) there _at least_ 
+                                     ///< should be so that the caller can succesfully insert the 
+                                     ///< desired samples to the buffer. If necessary, the function 
+                                     ///< grows the buffer size to comply with this requirement.
+                );
+
+    /// Adds 'numSamples' pcs of samples from the 'samples' memory position to
+    /// the sample buffer.
+    virtual void putSamples(const SAMPLETYPE *samples,  ///< Pointer to samples.
+                            uint numSamples                         ///< Number of samples to insert.
+                            );
+
+    /// Adjusts the book-keeping to increase number of samples in the buffer without 
+    /// copying any actual samples.
+    ///
+    /// This function is used to update the number of samples in the sample buffer
+    /// when accessing the buffer directly with 'ptrEnd' function. Please be 
+    /// careful though!
+    virtual void putSamples(uint numSamples   ///< Number of samples been inserted.
+                            );
+
+    /// Output samples from beginning of the sample buffer. Copies requested samples to 
+    /// output buffer and removes them from the sample buffer. If there are less than 
+    /// 'numsample' samples in the buffer, returns all that available.
+    ///
+    /// \return Number of samples returned.
+    virtual uint receiveSamples(SAMPLETYPE *output, ///< Buffer where to copy output samples.
+                                uint maxSamples                 ///< How many samples to receive at max.
+                                );
+
+    /// Adjusts book-keeping so that given number of samples are removed from beginning of the 
+    /// sample buffer without copying them anywhere. 
+    ///
+    /// Used to reduce the number of samples in the buffer when accessing the sample buffer directly
+    /// with 'ptrBegin' function.
+    virtual uint receiveSamples(uint maxSamples   ///< Remove this many samples from the beginning of pipe.
+                                );
+
+    /// Returns number of samples currently available.
+    virtual uint numSamples() const;
+
+    /// Sets number of channels, 1 = mono, 2 = stereo.
+    void setChannels(int numChannels);
+
+    /// Returns nonzero if there aren't any samples available for outputting.
+    virtual int isEmpty() const;
+
+    /// Clears all the samples.
+    virtual void clear();
+
+    /// allow trimming (downwards) amount of samples in pipeline.
+    /// Returns adjusted amount of samples
+    uint adjustAmountOfSamples(uint numSamples);
+};
+
+}
+
+#endif

Externals/soundtouch/FIFOSamplePipe.h

@@ -0,0 +1,234 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// 'FIFOSamplePipe' : An abstract base class for classes that manipulate sound
+/// samples by operating like a first-in-first-out pipe: New samples are fed
+/// into one end of the pipe with the 'putSamples' function, and the processed
+/// samples are received from the other end with the 'receiveSamples' function.
+///
+/// 'FIFOProcessor' : A base class for classes the do signal processing with 
+/// the samples while operating like a first-in-first-out pipe. When samples
+/// are input with the 'putSamples' function, the class processes them
+/// and moves the processed samples to the given 'output' pipe object, which
+/// may be either another processing stage, or a fifo sample buffer object.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-06-13 22:29:53 +0300 (Wed, 13 Jun 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: FIFOSamplePipe.h 143 2012-06-13 19:29:53Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef FIFOSamplePipe_H
+#define FIFOSamplePipe_H
+
+#include <assert.h>
+#include <stdlib.h>
+#include "STTypes.h"
+
+namespace soundtouch
+{
+
+/// Abstract base class for FIFO (first-in-first-out) sample processing classes.
+class FIFOSamplePipe
+{
+public:
+    // virtual default destructor
+    virtual ~FIFOSamplePipe() {}
+
+
+    /// Returns a pointer to the beginning of the output samples. 
+    /// This function is provided for accessing the output samples directly. 
+    /// Please be careful for not to corrupt the book-keeping!
+    ///
+    /// When using this function to output samples, also remember to 'remove' the
+    /// output samples from the buffer by calling the 
+    /// 'receiveSamples(numSamples)' function
+    virtual SAMPLETYPE *ptrBegin() = 0;
+
+    /// Adds 'numSamples' pcs of samples from the 'samples' memory position to
+    /// the sample buffer.
+    virtual void putSamples(const SAMPLETYPE *samples,  ///< Pointer to samples.
+                            uint numSamples             ///< Number of samples to insert.
+                            ) = 0;
+
+
+    // Moves samples from the 'other' pipe instance to this instance.
+    void moveSamples(FIFOSamplePipe &other  ///< Other pipe instance where from the receive the data.
+         )
+    {
+        int oNumSamples = other.numSamples();
+
+        putSamples(other.ptrBegin(), oNumSamples);
+        other.receiveSamples(oNumSamples);
+    };
+
+    /// Output samples from beginning of the sample buffer. Copies requested samples to 
+    /// output buffer and removes them from the sample buffer. If there are less than 
+    /// 'numsample' samples in the buffer, returns all that available.
+    ///
+    /// \return Number of samples returned.
+    virtual uint receiveSamples(SAMPLETYPE *output, ///< Buffer where to copy output samples.
+                                uint maxSamples                 ///< How many samples to receive at max.
+                                ) = 0;
+
+    /// Adjusts book-keeping so that given number of samples are removed from beginning of the 
+    /// sample buffer without copying them anywhere. 
+    ///
+    /// Used to reduce the number of samples in the buffer when accessing the sample buffer directly
+    /// with 'ptrBegin' function.
+    virtual uint receiveSamples(uint maxSamples   ///< Remove this many samples from the beginning of pipe.
+                                ) = 0;
+
+    /// Returns number of samples currently available.
+    virtual uint numSamples() const = 0;
+
+    // Returns nonzero if there aren't any samples available for outputting.
+    virtual int isEmpty() const = 0;
+
+    /// Clears all the samples.
+    virtual void clear() = 0;
+
+    /// allow trimming (downwards) amount of samples in pipeline.
+    /// Returns adjusted amount of samples
+    virtual uint adjustAmountOfSamples(uint numSamples) = 0;
+
+};
+
+
+
+/// Base-class for sound processing routines working in FIFO principle. With this base 
+/// class it's easy to implement sound processing stages that can be chained together,
+/// so that samples that are fed into beginning of the pipe automatically go through 
+/// all the processing stages.
+///
+/// When samples are input to this class, they're first processed and then put to 
+/// the FIFO pipe that's defined as output of this class. This output pipe can be
+/// either other processing stage or a FIFO sample buffer.
+class FIFOProcessor :public FIFOSamplePipe
+{
+protected:
+    /// Internal pipe where processed samples are put.
+    FIFOSamplePipe *output;
+
+    /// Sets output pipe.
+    void setOutPipe(FIFOSamplePipe *pOutput)
+    {
+        assert(output == NULL);
+        assert(pOutput != NULL);
+        output = pOutput;
+    }
+
+
+    /// Constructor. Doesn't define output pipe; it has to be set be 
+    /// 'setOutPipe' function.
+    FIFOProcessor()
+    {
+        output = NULL;
+    }
+
+
+    /// Constructor. Configures output pipe.
+    FIFOProcessor(FIFOSamplePipe *pOutput   ///< Output pipe.
+                 )
+    {
+        output = pOutput;
+    }
+
+
+    /// Destructor.
+    virtual ~FIFOProcessor()
+    {
+    }
+
+
+    /// Returns a pointer to the beginning of the output samples. 
+    /// This function is provided for accessing the output samples directly. 
+    /// Please be careful for not to corrupt the book-keeping!
+    ///
+    /// When using this function to output samples, also remember to 'remove' the
+    /// output samples from the buffer by calling the 
+    /// 'receiveSamples(numSamples)' function
+    virtual SAMPLETYPE *ptrBegin()
+    {
+        return output->ptrBegin();
+    }
+
+public:
+
+    /// Output samples from beginning of the sample buffer. Copies requested samples to 
+    /// output buffer and removes them from the sample buffer. If there are less than 
+    /// 'numsample' samples in the buffer, returns all that available.
+    ///
+    /// \return Number of samples returned.
+    virtual uint receiveSamples(SAMPLETYPE *outBuffer, ///< Buffer where to copy output samples.
+                                uint maxSamples                    ///< How many samples to receive at max.
+                                )
+    {
+        return output->receiveSamples(outBuffer, maxSamples);
+    }
+
+
+    /// Adjusts book-keeping so that given number of samples are removed from beginning of the 
+    /// sample buffer without copying them anywhere. 
+    ///
+    /// Used to reduce the number of samples in the buffer when accessing the sample buffer directly
+    /// with 'ptrBegin' function.
+    virtual uint receiveSamples(uint maxSamples   ///< Remove this many samples from the beginning of pipe.
+                                )
+    {
+        return output->receiveSamples(maxSamples);
+    }
+
+
+    /// Returns number of samples currently available.
+    virtual uint numSamples() const
+    {
+        return output->numSamples();
+    }
+
+
+    /// Returns nonzero if there aren't any samples available for outputting.
+    virtual int isEmpty() const
+    {
+        return output->isEmpty();
+    }
+
+    /// allow trimming (downwards) amount of samples in pipeline.
+    /// Returns adjusted amount of samples
+    virtual uint adjustAmountOfSamples(uint numSamples)
+    {
+        return output->adjustAmountOfSamples(numSamples);
+    }
+
+};
+
+}
+
+#endif

Externals/soundtouch/FIRFilter.cpp

@@ -0,0 +1,259 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// General FIR digital filter routines with MMX optimization. 
+///
+/// Note : MMX optimized functions reside in a separate, platform-specific file, 
+/// e.g. 'mmx_win.cpp' or 'mmx_gcc.cpp'
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2011-09-02 21:56:11 +0300 (Fri, 02 Sep 2011) $
+// File revision : $Revision: 4 $
+//
+// $Id: FIRFilter.cpp 131 2011-09-02 18:56:11Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <memory.h>
+#include <assert.h>
+#include <math.h>
+#include <stdlib.h>
+#include "FIRFilter.h"
+#include "cpu_detect.h"
+
+using namespace soundtouch;
+
+/*****************************************************************************
+ *
+ * Implementation of the class 'FIRFilter'
+ *
+ *****************************************************************************/
+
+FIRFilter::FIRFilter()
+{
+    resultDivFactor = 0;
+    resultDivider = 0;
+    length = 0;
+    lengthDiv8 = 0;
+    filterCoeffs = NULL;
+}
+
+
+FIRFilter::~FIRFilter()
+{
+    delete[] filterCoeffs;
+}
+
+// Usual C-version of the filter routine for stereo sound
+uint FIRFilter::evaluateFilterStereo(SAMPLETYPE *dest, const SAMPLETYPE *src, uint numSamples) const
+{
+    uint i, j, end;
+    LONG_SAMPLETYPE suml, sumr;
+#ifdef SOUNDTOUCH_FLOAT_SAMPLES
+    // when using floating point samples, use a scaler instead of a divider
+    // because division is much slower operation than multiplying.
+    double dScaler = 1.0 / (double)resultDivider;
+#endif
+
+    assert(length != 0);
+    assert(src != NULL);
+    assert(dest != NULL);
+    assert(filterCoeffs != NULL);
+
+    end = 2 * (numSamples - length);
+
+    for (j = 0; j < end; j += 2) 
+    {
+        const SAMPLETYPE *ptr;
+
+        suml = sumr = 0;
+        ptr = src + j;
+
+        for (i = 0; i < length; i += 4) 
+        {
+            // loop is unrolled by factor of 4 here for efficiency
+            suml += ptr[2 * i + 0] * filterCoeffs[i + 0] +
+                    ptr[2 * i + 2] * filterCoeffs[i + 1] +
+                    ptr[2 * i + 4] * filterCoeffs[i + 2] +
+                    ptr[2 * i + 6] * filterCoeffs[i + 3];
+            sumr += ptr[2 * i + 1] * filterCoeffs[i + 0] +
+                    ptr[2 * i + 3] * filterCoeffs[i + 1] +
+                    ptr[2 * i + 5] * filterCoeffs[i + 2] +
+                    ptr[2 * i + 7] * filterCoeffs[i + 3];
+        }
+
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+        suml >>= resultDivFactor;
+        sumr >>= resultDivFactor;
+        // saturate to 16 bit integer limits
+        suml = (suml < -32768) ? -32768 : (suml > 32767) ? 32767 : suml;
+        // saturate to 16 bit integer limits
+        sumr = (sumr < -32768) ? -32768 : (sumr > 32767) ? 32767 : sumr;
+#else
+        suml *= dScaler;
+        sumr *= dScaler;
+#endif // SOUNDTOUCH_INTEGER_SAMPLES
+        dest[j] = (SAMPLETYPE)suml;
+        dest[j + 1] = (SAMPLETYPE)sumr;
+    }
+    return numSamples - length;
+}
+
+
+
+
+// Usual C-version of the filter routine for mono sound
+uint FIRFilter::evaluateFilterMono(SAMPLETYPE *dest, const SAMPLETYPE *src, uint numSamples) const
+{
+    uint i, j, end;
+    LONG_SAMPLETYPE sum;
+#ifdef SOUNDTOUCH_FLOAT_SAMPLES
+    // when using floating point samples, use a scaler instead of a divider
+    // because division is much slower operation than multiplying.
+    double dScaler = 1.0 / (double)resultDivider;
+#endif
+
+
+    assert(length != 0);
+
+    end = numSamples - length;
+    for (j = 0; j < end; j ++) 
+    {
+        sum = 0;
+        for (i = 0; i < length; i += 4) 
+        {
+            // loop is unrolled by factor of 4 here for efficiency
+            sum += src[i + 0] * filterCoeffs[i + 0] + 
+                   src[i + 1] * filterCoeffs[i + 1] + 
+                   src[i + 2] * filterCoeffs[i + 2] + 
+                   src[i + 3] * filterCoeffs[i + 3];
+        }
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+        sum >>= resultDivFactor;
+        // saturate to 16 bit integer limits
+        sum = (sum < -32768) ? -32768 : (sum > 32767) ? 32767 : sum;
+#else
+        sum *= dScaler;
+#endif // SOUNDTOUCH_INTEGER_SAMPLES
+        dest[j] = (SAMPLETYPE)sum;
+        src ++;
+    }
+    return end;
+}
+
+
+// Set filter coeffiecients and length.
+//
+// Throws an exception if filter length isn't divisible by 8
+void FIRFilter::setCoefficients(const SAMPLETYPE *coeffs, uint newLength, uint uResultDivFactor)
+{
+    assert(newLength > 0);
+    if (newLength % 8) ST_THROW_RT_ERROR("FIR filter length not divisible by 8");
+
+    lengthDiv8 = newLength / 8;
+    length = lengthDiv8 * 8;
+    assert(length == newLength);
+
+    resultDivFactor = uResultDivFactor;
+    resultDivider = (SAMPLETYPE)::pow(2.0, (int)resultDivFactor);
+
+    delete[] filterCoeffs;
+    filterCoeffs = new SAMPLETYPE[length];
+    memcpy(filterCoeffs, coeffs, length * sizeof(SAMPLETYPE));
+}
+
+
+uint FIRFilter::getLength() const
+{
+    return length;
+}
+
+
+
+// Applies the filter to the given sequence of samples. 
+//
+// Note : The amount of outputted samples is by value of 'filter_length' 
+// smaller than the amount of input samples.
+uint FIRFilter::evaluate(SAMPLETYPE *dest, const SAMPLETYPE *src, uint numSamples, uint numChannels) const
+{
+    assert(numChannels == 1 || numChannels == 2);
+
+    assert(length > 0);
+    assert(lengthDiv8 * 8 == length);
+    if (numSamples < length) return 0;
+    if (numChannels == 2) 
+    {
+        return evaluateFilterStereo(dest, src, numSamples);
+    } else {
+        return evaluateFilterMono(dest, src, numSamples);
+    }
+}
+
+
+
+// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+// depending on if we've a MMX-capable CPU available or not.
+void * FIRFilter::operator new(size_t s)
+{
+    // Notice! don't use "new FIRFilter" directly, use "newInstance" to create a new instance instead!
+    ST_THROW_RT_ERROR("Error in FIRFilter::new: Don't use 'new FIRFilter', use 'newInstance' member instead!");
+    return newInstance();
+}
+
+
+FIRFilter * FIRFilter::newInstance()
+{
+    uint uExtensions;
+
+    uExtensions = detectCPUextensions();
+
+    // Check if MMX/SSE instruction set extensions supported by CPU
+
+#ifdef SOUNDTOUCH_ALLOW_MMX
+    // MMX routines available only with integer sample types
+    if (uExtensions & SUPPORT_MMX)
+    {
+        return ::new FIRFilterMMX;
+    }
+    else
+#endif // SOUNDTOUCH_ALLOW_MMX
+
+#ifdef SOUNDTOUCH_ALLOW_SSE
+    if (uExtensions & SUPPORT_SSE)
+    {
+        // SSE support
+        return ::new FIRFilterSSE;
+    }
+    else
+#endif // SOUNDTOUCH_ALLOW_SSE
+
+    {
+        // ISA optimizations not supported, use plain C version
+        return ::new FIRFilter;
+    }
+}

Externals/soundtouch/FIRFilter.h

@@ -0,0 +1,145 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// General FIR digital filter routines with MMX optimization. 
+///
+/// Note : MMX optimized functions reside in a separate, platform-specific file, 
+/// e.g. 'mmx_win.cpp' or 'mmx_gcc.cpp'
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2011-02-13 21:13:57 +0200 (Sun, 13 Feb 2011) $
+// File revision : $Revision: 4 $
+//
+// $Id: FIRFilter.h 104 2011-02-13 19:13:57Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef FIRFilter_H
+#define FIRFilter_H
+
+#include <stddef.h>
+#include "STTypes.h"
+
+namespace soundtouch
+{
+
+class FIRFilter 
+{
+protected:
+    // Number of FIR filter taps
+    uint length;    
+    // Number of FIR filter taps divided by 8
+    uint lengthDiv8;
+
+    // Result divider factor in 2^k format
+    uint resultDivFactor;
+
+    // Result divider value.
+    SAMPLETYPE resultDivider;
+
+    // Memory for filter coefficients
+    SAMPLETYPE *filterCoeffs;
+
+    virtual uint evaluateFilterStereo(SAMPLETYPE *dest, 
+                                      const SAMPLETYPE *src, 
+                                      uint numSamples) const;
+    virtual uint evaluateFilterMono(SAMPLETYPE *dest, 
+                                    const SAMPLETYPE *src, 
+                                    uint numSamples) const;
+
+public:
+    FIRFilter();
+    virtual ~FIRFilter();
+
+    /// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+    /// depending on if we've a MMX-capable CPU available or not.
+    static void * operator new(size_t s);
+
+    static FIRFilter *newInstance();
+
+    /// Applies the filter to the given sequence of samples. 
+    /// Note : The amount of outputted samples is by value of 'filter_length' 
+    /// smaller than the amount of input samples.
+    ///
+    /// \return Number of samples copied to 'dest'.
+    uint evaluate(SAMPLETYPE *dest, 
+                  const SAMPLETYPE *src, 
+                  uint numSamples, 
+                  uint numChannels) const;
+
+    uint getLength() const;
+
+    virtual void setCoefficients(const SAMPLETYPE *coeffs, 
+                                 uint newLength, 
+                                 uint uResultDivFactor);
+};
+
+
+// Optional subclasses that implement CPU-specific optimizations:
+
+#ifdef SOUNDTOUCH_ALLOW_MMX
+
+/// Class that implements MMX optimized functions exclusive for 16bit integer samples type.
+    class FIRFilterMMX : public FIRFilter
+    {
+    protected:
+        short *filterCoeffsUnalign;
+        short *filterCoeffsAlign;
+
+        virtual uint evaluateFilterStereo(short *dest, const short *src, uint numSamples) const;
+    public:
+        FIRFilterMMX();
+        ~FIRFilterMMX();
+
+        virtual void setCoefficients(const short *coeffs, uint newLength, uint uResultDivFactor);
+    };
+
+#endif // SOUNDTOUCH_ALLOW_MMX
+
+
+#ifdef SOUNDTOUCH_ALLOW_SSE
+    /// Class that implements SSE optimized functions exclusive for floating point samples type.
+    class FIRFilterSSE : public FIRFilter
+    {
+    protected:
+        float *filterCoeffsUnalign;
+        float *filterCoeffsAlign;
+
+        virtual uint evaluateFilterStereo(float *dest, const float *src, uint numSamples) const;
+    public:
+        FIRFilterSSE();
+        ~FIRFilterSSE();
+
+        virtual void setCoefficients(const float *coeffs, uint newLength, uint uResultDivFactor);
+    };
+
+#endif // SOUNDTOUCH_ALLOW_SSE
+
+}
+
+#endif  // FIRFilter_H

Externals/soundtouch/PeakFinder.cpp

@@ -0,0 +1,276 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Peak detection routine. 
+///
+/// The routine detects highest value on an array of values and calculates the 
+/// precise peak location as a mass-center of the 'hump' around the peak value.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-12-28 21:52:47 +0200 (Fri, 28 Dec 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: PeakFinder.cpp 164 2012-12-28 19:52:47Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <math.h>
+#include <assert.h>
+
+#include "PeakFinder.h"
+
+using namespace soundtouch;
+
+#define max(x, y) (((x) > (y)) ? (x) : (y))
+
+
+PeakFinder::PeakFinder()
+{
+    minPos = maxPos = 0;
+}
+
+
+// Finds real 'top' of a peak hump from neighnourhood of the given 'peakpos'.
+int PeakFinder::findTop(const float *data, int peakpos) const
+{
+    int i;
+    int start, end;
+    float refvalue;
+
+    refvalue = data[peakpos];
+
+    // seek within <20>10 points
+    start = peakpos - 10;
+    if (start < minPos) start = minPos;
+    end = peakpos + 10;
+    if (end > maxPos) end = maxPos;
+
+    for (i = start; i <= end; i ++)
+    {
+        if (data[i] > refvalue)
+        {
+            peakpos = i;
+            refvalue = data[i];
+        }
+    }
+
+    // failure if max value is at edges of seek range => it's not peak, it's at slope.
+    if ((peakpos == start) || (peakpos == end)) return 0;
+
+    return peakpos;
+}
+
+
+// Finds 'ground level' of a peak hump by starting from 'peakpos' and proceeding
+// to direction defined by 'direction' until next 'hump' after minimum value will 
+// begin
+int PeakFinder::findGround(const float *data, int peakpos, int direction) const
+{
+    int lowpos;
+    int pos;
+    int climb_count;
+    float refvalue;
+    float delta;
+
+    climb_count = 0;
+    refvalue = data[peakpos];
+    lowpos = peakpos;
+
+    pos = peakpos;
+
+    while ((pos > minPos+1) && (pos < maxPos-1))
+    {
+        int prevpos;
+
+        prevpos = pos;
+        pos += direction;
+
+        // calculate derivate
+        delta = data[pos] - data[prevpos];
+        if (delta <= 0)
+        {
+            // going downhill, ok
+            if (climb_count)
+            {
+                climb_count --;  // decrease climb count
+            }
+
+            // check if new minimum found
+            if (data[pos] < refvalue)
+            {
+                // new minimum found
+                lowpos = pos;
+                refvalue = data[pos];
+            }
+        }
+        else
+        {
+            // going uphill, increase climbing counter
+            climb_count ++;
+            if (climb_count > 5) break;    // we've been climbing too long => it's next uphill => quit
+        }
+    }
+    return lowpos;
+}
+
+
+// Find offset where the value crosses the given level, when starting from 'peakpos' and
+// proceeds to direction defined in 'direction'
+int PeakFinder::findCrossingLevel(const float *data, float level, int peakpos, int direction) const
+{
+    float peaklevel;
+    int pos;
+
+    peaklevel = data[peakpos];
+    assert(peaklevel >= level);
+    pos = peakpos;
+    while ((pos >= minPos) && (pos < maxPos))
+    {
+        if (data[pos + direction] < level) return pos;   // crossing found
+        pos += direction;
+    }
+    return -1;  // not found
+}
+
+
+// Calculates the center of mass location of 'data' array items between 'firstPos' and 'lastPos'
+double PeakFinder::calcMassCenter(const float *data, int firstPos, int lastPos) const
+{
+    int i;
+    float sum;
+    float wsum;
+
+    sum = 0;
+    wsum = 0;
+    for (i = firstPos; i <= lastPos; i ++)
+    {
+        sum += (float)i * data[i];
+        wsum += data[i];
+    }
+
+    if (wsum < 1e-6) return 0;
+    return sum / wsum;
+}
+
+
+
+/// get exact center of peak near given position by calculating local mass of center
+double PeakFinder::getPeakCenter(const float *data, int peakpos) const
+{
+    float peakLevel;            // peak level
+    int crosspos1, crosspos2;   // position where the peak 'hump' crosses cutting level
+    float cutLevel;             // cutting value
+    float groundLevel;          // ground level of the peak
+    int gp1, gp2;               // bottom positions of the peak 'hump'
+
+    // find ground positions.
+    gp1 = findGround(data, peakpos, -1);
+    gp2 = findGround(data, peakpos, 1);
+
+    groundLevel = 0.5f * (data[gp1] + data[gp2]);
+    peakLevel = data[peakpos];
+
+    // calculate 70%-level of the peak
+    cutLevel = 0.70f * peakLevel + 0.30f * groundLevel;
+    // find mid-level crossings
+    crosspos1 = findCrossingLevel(data, cutLevel, peakpos, -1);
+    crosspos2 = findCrossingLevel(data, cutLevel, peakpos, 1);
+
+    if ((crosspos1 < 0) || (crosspos2 < 0)) return 0;   // no crossing, no peak..
+
+    // calculate mass center of the peak surroundings
+    return calcMassCenter(data, crosspos1, crosspos2);
+}
+
+
+
+double PeakFinder::detectPeak(const float *data, int aminPos, int amaxPos) 
+{
+
+    int i;
+    int peakpos;                // position of peak level
+    double highPeak, peak;
+
+    this->minPos = aminPos;
+    this->maxPos = amaxPos;
+
+    // find absolute peak
+    peakpos = minPos;
+    peak = data[minPos];
+    for (i = minPos + 1; i < maxPos; i ++)
+    {
+        if (data[i] > peak) 
+        {
+            peak = data[i];
+            peakpos = i;
+        }
+    }
+    
+    // Calculate exact location of the highest peak mass center
+    highPeak = getPeakCenter(data, peakpos);
+    peak = highPeak;
+
+    // Now check if the highest peak were in fact harmonic of the true base beat peak 
+    // - sometimes the highest peak can be Nth harmonic of the true base peak yet 
+    // just a slightly higher than the true base
+
+    for (i = 3; i < 10; i ++)
+    {
+        double peaktmp, harmonic;
+        int i1,i2;
+
+        harmonic = (double)i * 0.5;
+        peakpos = (int)(highPeak / harmonic + 0.5f);
+        if (peakpos < minPos) break;
+        peakpos = findTop(data, peakpos);   // seek true local maximum index
+        if (peakpos == 0) continue;         // no local max here
+
+        // calculate mass-center of possible harmonic peak
+        peaktmp = getPeakCenter(data, peakpos);
+
+        // accept harmonic peak if 
+        // (a) it is found
+        // (b) is within <20>4% of the expected harmonic interval
+        // (c) has at least half x-corr value of the max. peak
+
+        double diff = harmonic * peaktmp / highPeak;
+        if ((diff < 0.96) || (diff > 1.04)) continue;   // peak too afar from expected
+
+        // now compare to highest detected peak
+        i1 = (int)(highPeak + 0.5);
+        i2 = (int)(peaktmp + 0.5);
+        if (data[i2] >= 0.4*data[i1])
+        {
+            // The harmonic is at least half as high primary peak,
+            // thus use the harmonic peak instead
+            peak = peaktmp;
+        }
+    }
+
+    return peak;
+}

Externals/soundtouch/PeakFinder.h

@@ -0,0 +1,97 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// The routine detects highest value on an array of values and calculates the 
+/// precise peak location as a mass-center of the 'hump' around the peak value.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2011-12-30 22:33:46 +0200 (Fri, 30 Dec 2011) $
+// File revision : $Revision: 4 $
+//
+// $Id: PeakFinder.h 132 2011-12-30 20:33:46Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef _PeakFinder_H_
+#define _PeakFinder_H_
+
+namespace soundtouch
+{
+
+class PeakFinder
+{
+protected:
+    /// Min, max allowed peak positions within the data vector
+    int minPos, maxPos;
+
+    /// Calculates the mass center between given vector items.
+    double calcMassCenter(const float *data, ///< Data vector.
+                         int firstPos,      ///< Index of first vector item beloging to the peak.
+                         int lastPos        ///< Index of last vector item beloging to the peak.
+                         ) const;
+
+    /// Finds the data vector index where the monotoniously decreasing signal crosses the
+    /// given level.
+    int   findCrossingLevel(const float *data,  ///< Data vector.
+                            float level,        ///< Goal crossing level.
+                            int peakpos,        ///< Peak position index within the data vector.
+                            int direction       /// Direction where to proceed from the peak: 1 = right, -1 = left.
+                            ) const;
+
+    // Finds real 'top' of a peak hump from neighnourhood of the given 'peakpos'.
+    int findTop(const float *data, int peakpos) const;
+
+
+    /// Finds the 'ground' level, i.e. smallest level between two neighbouring peaks, to right- 
+    /// or left-hand side of the given peak position.
+    int   findGround(const float *data,     /// Data vector.
+                     int peakpos,           /// Peak position index within the data vector.
+                     int direction          /// Direction where to proceed from the peak: 1 = right, -1 = left.
+                     ) const;
+
+    /// get exact center of peak near given position by calculating local mass of center
+    double getPeakCenter(const float *data, int peakpos) const;
+
+public:
+    /// Constructor. 
+    PeakFinder();
+
+    /// Detect exact peak position of the data vector by finding the largest peak 'hump'
+    /// and calculating the mass-center location of the peak hump.
+    ///
+    /// \return The location of the largest base harmonic peak hump.
+    double detectPeak(const float *data, /// Data vector to be analyzed. The data vector has
+                                        /// to be at least 'maxPos' items long.
+                     int minPos,        ///< Min allowed peak location within the vector data.
+                     int maxPos         ///< Max allowed peak location within the vector data.
+                     );
+};
+
+}
+
+#endif // _PeakFinder_H_

Externals/soundtouch/RateTransposer.cpp

@@ -0,0 +1,626 @@
+////////////////////////////////////////////////////////////////////////////////
+/// 
+/// Sample rate transposer. Changes sample rate by using linear interpolation 
+/// together with anti-alias filtering (first order interpolation with anti-
+/// alias filtering should be quite adequate for this application)
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2011-09-02 21:56:11 +0300 (Fri, 02 Sep 2011) $
+// File revision : $Revision: 4 $
+//
+// $Id: RateTransposer.cpp 131 2011-09-02 18:56:11Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <memory.h>
+#include <assert.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include "RateTransposer.h"
+#include "AAFilter.h"
+
+using namespace soundtouch;
+
+
+/// A linear samplerate transposer class that uses integer arithmetics.
+/// for the transposing.
+class RateTransposerInteger : public RateTransposer
+{
+protected:
+    int iSlopeCount;
+    int iRate;
+    SAMPLETYPE sPrevSampleL, sPrevSampleR;
+
+    virtual void resetRegisters();
+
+    virtual uint transposeStereo(SAMPLETYPE *dest, 
+                         const SAMPLETYPE *src, 
+                         uint numSamples);
+    virtual uint transposeMono(SAMPLETYPE *dest, 
+                       const SAMPLETYPE *src, 
+                       uint numSamples);
+
+public:
+    RateTransposerInteger();
+    virtual ~RateTransposerInteger();
+
+    /// Sets new target rate. Normal rate = 1.0, smaller values represent slower 
+    /// rate, larger faster rates.
+    virtual void setRate(float newRate);
+
+};
+
+
+/// A linear samplerate transposer class that uses floating point arithmetics
+/// for the transposing.
+class RateTransposerFloat : public RateTransposer
+{
+protected:
+    float fSlopeCount;
+    SAMPLETYPE sPrevSampleL, sPrevSampleR;
+
+    virtual void resetRegisters();
+
+    virtual uint transposeStereo(SAMPLETYPE *dest, 
+                         const SAMPLETYPE *src, 
+                         uint numSamples);
+    virtual uint transposeMono(SAMPLETYPE *dest, 
+                       const SAMPLETYPE *src, 
+                       uint numSamples);
+
+public:
+    RateTransposerFloat();
+    virtual ~RateTransposerFloat();
+};
+
+
+
+
+// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+// depending on if we've a MMX/SSE/etc-capable CPU available or not.
+void * RateTransposer::operator new(size_t s)
+{
+    ST_THROW_RT_ERROR("Error in RateTransoser::new: don't use \"new TDStretch\" directly, use \"newInstance\" to create a new instance instead!");
+    return newInstance();
+}
+
+
+RateTransposer *RateTransposer::newInstance()
+{
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+    return ::new RateTransposerInteger;
+#else
+    return ::new RateTransposerFloat;
+#endif
+}
+
+
+// Constructor
+RateTransposer::RateTransposer() : FIFOProcessor(&outputBuffer)
+{
+    numChannels = 2;
+    bUseAAFilter = TRUE;
+    fRate = 0;
+
+    // Instantiates the anti-alias filter with default tap length
+    // of 32
+    pAAFilter = new AAFilter(32);
+}
+
+
+
+RateTransposer::~RateTransposer()
+{
+    delete pAAFilter;
+}
+
+
+
+/// Enables/disables the anti-alias filter. Zero to disable, nonzero to enable
+void RateTransposer::enableAAFilter(BOOL newMode)
+{
+    bUseAAFilter = newMode;
+}
+
+
+/// Returns nonzero if anti-alias filter is enabled.
+BOOL RateTransposer::isAAFilterEnabled() const
+{
+    return bUseAAFilter;
+}
+
+
+AAFilter *RateTransposer::getAAFilter()
+{
+    return pAAFilter;
+}
+
+
+
+// Sets new target iRate. Normal iRate = 1.0, smaller values represent slower 
+// iRate, larger faster iRates.
+void RateTransposer::setRate(float newRate)
+{
+    double fCutoff;
+
+    fRate = newRate;
+
+    // design a new anti-alias filter
+    if (newRate > 1.0f) 
+    {
+        fCutoff = 0.5f / newRate;
+    } 
+    else 
+    {
+        fCutoff = 0.5f * newRate;
+    }
+    pAAFilter->setCutoffFreq(fCutoff);
+}
+
+
+// Outputs as many samples of the 'outputBuffer' as possible, and if there's
+// any room left, outputs also as many of the incoming samples as possible.
+// The goal is to drive the outputBuffer empty.
+//
+// It's allowed for 'output' and 'input' parameters to point to the same
+// memory position.
+/*
+void RateTransposer::flushStoreBuffer()
+{
+    if (storeBuffer.isEmpty()) return;
+
+    outputBuffer.moveSamples(storeBuffer);
+}
+*/
+
+
+// Adds 'nSamples' pcs of samples from the 'samples' memory position into
+// the input of the object.
+void RateTransposer::putSamples(const SAMPLETYPE *samples, uint nSamples)
+{
+    processSamples(samples, nSamples);
+}
+
+
+
+// Transposes up the sample rate, causing the observed playback 'rate' of the
+// sound to decrease
+void RateTransposer::upsample(const SAMPLETYPE *src, uint nSamples)
+{
+    uint count, sizeTemp, num;
+
+    // If the parameter 'uRate' value is smaller than 'SCALE', first transpose
+    // the samples and then apply the anti-alias filter to remove aliasing.
+
+    // First check that there's enough room in 'storeBuffer' 
+    // (+16 is to reserve some slack in the destination buffer)
+    sizeTemp = (uint)((float)nSamples / fRate + 16.0f);
+
+    // Transpose the samples, store the result into the end of "storeBuffer"
+    count = transpose(storeBuffer.ptrEnd(sizeTemp), src, nSamples);
+    storeBuffer.putSamples(count);
+
+    // Apply the anti-alias filter to samples in "store output", output the
+    // result to "dest"
+    num = storeBuffer.numSamples();
+    count = pAAFilter->evaluate(outputBuffer.ptrEnd(num), 
+        storeBuffer.ptrBegin(), num, (uint)numChannels);
+    outputBuffer.putSamples(count);
+
+    // Remove the processed samples from "storeBuffer"
+    storeBuffer.receiveSamples(count);
+}
+
+
+// Transposes down the sample rate, causing the observed playback 'rate' of the
+// sound to increase
+void RateTransposer::downsample(const SAMPLETYPE *src, uint nSamples)
+{
+    uint count, sizeTemp;
+
+    // If the parameter 'uRate' value is larger than 'SCALE', first apply the
+    // anti-alias filter to remove high frequencies (prevent them from folding
+    // over the lover frequencies), then transpose.
+
+    // Add the new samples to the end of the storeBuffer
+    storeBuffer.putSamples(src, nSamples);
+
+    // Anti-alias filter the samples to prevent folding and output the filtered 
+    // data to tempBuffer. Note : because of the FIR filter length, the
+    // filtering routine takes in 'filter_length' more samples than it outputs.
+    assert(tempBuffer.isEmpty());
+    sizeTemp = storeBuffer.numSamples();
+
+    count = pAAFilter->evaluate(tempBuffer.ptrEnd(sizeTemp), 
+        storeBuffer.ptrBegin(), sizeTemp, (uint)numChannels);
+
+	if (count == 0) return;
+
+    // Remove the filtered samples from 'storeBuffer'
+    storeBuffer.receiveSamples(count);
+
+    // Transpose the samples (+16 is to reserve some slack in the destination buffer)
+    sizeTemp = (uint)((float)nSamples / fRate + 16.0f);
+    count = transpose(outputBuffer.ptrEnd(sizeTemp), tempBuffer.ptrBegin(), count);
+    outputBuffer.putSamples(count);
+}
+
+
+// Transposes sample rate by applying anti-alias filter to prevent folding. 
+// Returns amount of samples returned in the "dest" buffer.
+// The maximum amount of samples that can be returned at a time is set by
+// the 'set_returnBuffer_size' function.
+void RateTransposer::processSamples(const SAMPLETYPE *src, uint nSamples)
+{
+    uint count;
+    uint sizeReq;
+
+    if (nSamples == 0) return;
+    assert(pAAFilter);
+
+    // If anti-alias filter is turned off, simply transpose without applying
+    // the filter
+    if (bUseAAFilter == FALSE) 
+    {
+        sizeReq = (uint)((float)nSamples / fRate + 1.0f);
+        count = transpose(outputBuffer.ptrEnd(sizeReq), src, nSamples);
+        outputBuffer.putSamples(count);
+        return;
+    }
+
+    // Transpose with anti-alias filter
+    if (fRate < 1.0f) 
+    {
+        upsample(src, nSamples);
+    } 
+    else  
+    {
+        downsample(src, nSamples);
+    }
+}
+
+
+// Transposes the sample rate of the given samples using linear interpolation. 
+// Returns the number of samples returned in the "dest" buffer
+inline uint RateTransposer::transpose(SAMPLETYPE *dest, const SAMPLETYPE *src, uint nSamples)
+{
+    if (numChannels == 2) 
+    {
+        return transposeStereo(dest, src, nSamples);
+    } 
+    else 
+    {
+        return transposeMono(dest, src, nSamples);
+    }
+}
+
+
+// Sets the number of channels, 1 = mono, 2 = stereo
+void RateTransposer::setChannels(int nChannels)
+{
+    assert(nChannels > 0);
+    if (numChannels == nChannels) return;
+
+    assert(nChannels == 1 || nChannels == 2);
+    numChannels = nChannels;
+
+    storeBuffer.setChannels(numChannels);
+    tempBuffer.setChannels(numChannels);
+    outputBuffer.setChannels(numChannels);
+
+    // Inits the linear interpolation registers
+    resetRegisters();
+}
+
+
+// Clears all the samples in the object
+void RateTransposer::clear()
+{
+    outputBuffer.clear();
+    storeBuffer.clear();
+}
+
+
+// Returns nonzero if there aren't any samples available for outputting.
+int RateTransposer::isEmpty() const
+{
+    int res;
+
+    res = FIFOProcessor::isEmpty();
+    if (res == 0) return 0;
+    return storeBuffer.isEmpty();
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// RateTransposerInteger - integer arithmetic implementation
+// 
+
+/// fixed-point interpolation routine precision
+#define SCALE    65536
+
+// Constructor
+RateTransposerInteger::RateTransposerInteger() : RateTransposer()
+{
+    // Notice: use local function calling syntax for sake of clarity, 
+    // to indicate the fact that C++ constructor can't call virtual functions.
+    RateTransposerInteger::resetRegisters();
+    RateTransposerInteger::setRate(1.0f);
+}
+
+
+RateTransposerInteger::~RateTransposerInteger()
+{
+}
+
+
+void RateTransposerInteger::resetRegisters()
+{
+    iSlopeCount = 0;
+    sPrevSampleL = 
+    sPrevSampleR = 0;
+}
+
+
+
+// Transposes the sample rate of the given samples using linear interpolation. 
+// 'Mono' version of the routine. Returns the number of samples returned in 
+// the "dest" buffer
+uint RateTransposerInteger::transposeMono(SAMPLETYPE *dest, const SAMPLETYPE *src, uint nSamples)
+{
+    unsigned int i, used;
+    LONG_SAMPLETYPE temp, vol1;
+
+    if (nSamples == 0) return 0;  // no samples, no work
+
+	used = 0;    
+    i = 0;
+
+    // Process the last sample saved from the previous call first...
+    while (iSlopeCount <= SCALE) 
+    {
+        vol1 = (LONG_SAMPLETYPE)(SCALE - iSlopeCount);
+        temp = vol1 * sPrevSampleL + iSlopeCount * src[0];
+        dest[i] = (SAMPLETYPE)(temp / SCALE);
+        i++;
+        iSlopeCount += iRate;
+    }
+    // now always (iSlopeCount > SCALE)
+    iSlopeCount -= SCALE;
+
+    while (1)
+    {
+        while (iSlopeCount > SCALE) 
+        {
+            iSlopeCount -= SCALE;
+            used ++;
+            if (used >= nSamples - 1) goto end;
+        }
+        vol1 = (LONG_SAMPLETYPE)(SCALE - iSlopeCount);
+        temp = src[used] * vol1 + iSlopeCount * src[used + 1];
+        dest[i] = (SAMPLETYPE)(temp / SCALE);
+
+        i++;
+        iSlopeCount += iRate;
+    }
+end:
+    // Store the last sample for the next round
+    sPrevSampleL = src[nSamples - 1];
+
+    return i;
+}
+
+
+// Transposes the sample rate of the given samples using linear interpolation. 
+// 'Stereo' version of the routine. Returns the number of samples returned in 
+// the "dest" buffer
+uint RateTransposerInteger::transposeStereo(SAMPLETYPE *dest, const SAMPLETYPE *src, uint nSamples)
+{
+    unsigned int srcPos, i, used;
+    LONG_SAMPLETYPE temp, vol1;
+
+    if (nSamples == 0) return 0;  // no samples, no work
+
+    used = 0;    
+    i = 0;
+
+    // Process the last sample saved from the sPrevSampleLious call first...
+    while (iSlopeCount <= SCALE) 
+    {
+        vol1 = (LONG_SAMPLETYPE)(SCALE - iSlopeCount);
+        temp = vol1 * sPrevSampleL + iSlopeCount * src[0];
+        dest[2 * i] = (SAMPLETYPE)(temp / SCALE);
+        temp = vol1 * sPrevSampleR + iSlopeCount * src[1];
+        dest[2 * i + 1] = (SAMPLETYPE)(temp / SCALE);
+        i++;
+        iSlopeCount += iRate;
+    }
+    // now always (iSlopeCount > SCALE)
+    iSlopeCount -= SCALE;
+
+    while (1)
+    {
+        while (iSlopeCount > SCALE) 
+        {
+            iSlopeCount -= SCALE;
+            used ++;
+            if (used >= nSamples - 1) goto end;
+        }
+        srcPos = 2 * used;
+        vol1 = (LONG_SAMPLETYPE)(SCALE - iSlopeCount);
+        temp = src[srcPos] * vol1 + iSlopeCount * src[srcPos + 2];
+        dest[2 * i] = (SAMPLETYPE)(temp / SCALE);
+        temp = src[srcPos + 1] * vol1 + iSlopeCount * src[srcPos + 3];
+        dest[2 * i + 1] = (SAMPLETYPE)(temp / SCALE);
+
+        i++;
+        iSlopeCount += iRate;
+    }
+end:
+    // Store the last sample for the next round
+    sPrevSampleL = src[2 * nSamples - 2];
+    sPrevSampleR = src[2 * nSamples - 1];
+
+    return i;
+}
+
+
+// Sets new target iRate. Normal iRate = 1.0, smaller values represent slower 
+// iRate, larger faster iRates.
+void RateTransposerInteger::setRate(float newRate)
+{
+    iRate = (int)(newRate * SCALE + 0.5f);
+    RateTransposer::setRate(newRate);
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// RateTransposerFloat - floating point arithmetic implementation
+// 
+//////////////////////////////////////////////////////////////////////////////
+
+// Constructor
+RateTransposerFloat::RateTransposerFloat() : RateTransposer()
+{
+    // Notice: use local function calling syntax for sake of clarity, 
+    // to indicate the fact that C++ constructor can't call virtual functions.
+    RateTransposerFloat::resetRegisters();
+    RateTransposerFloat::setRate(1.0f);
+}
+
+
+RateTransposerFloat::~RateTransposerFloat()
+{
+}
+
+
+void RateTransposerFloat::resetRegisters()
+{
+    fSlopeCount = 0;
+    sPrevSampleL = 
+    sPrevSampleR = 0;
+}
+
+
+
+// Transposes the sample rate of the given samples using linear interpolation. 
+// 'Mono' version of the routine. Returns the number of samples returned in 
+// the "dest" buffer
+uint RateTransposerFloat::transposeMono(SAMPLETYPE *dest, const SAMPLETYPE *src, uint nSamples)
+{
+    unsigned int i, used;
+
+    used = 0;    
+    i = 0;
+
+    // Process the last sample saved from the previous call first...
+    while (fSlopeCount <= 1.0f) 
+    {
+        dest[i] = (SAMPLETYPE)((1.0f - fSlopeCount) * sPrevSampleL + fSlopeCount * src[0]);
+        i++;
+        fSlopeCount += fRate;
+    }
+    fSlopeCount -= 1.0f;
+
+    if (nSamples > 1)
+    {
+        while (1)
+        {
+            while (fSlopeCount > 1.0f) 
+            {
+                fSlopeCount -= 1.0f;
+                used ++;
+                if (used >= nSamples - 1) goto end;
+            }
+            dest[i] = (SAMPLETYPE)((1.0f - fSlopeCount) * src[used] + fSlopeCount * src[used + 1]);
+            i++;
+            fSlopeCount += fRate;
+        }
+    }
+end:
+    // Store the last sample for the next round
+    sPrevSampleL = src[nSamples - 1];
+
+    return i;
+}
+
+
+// Transposes the sample rate of the given samples using linear interpolation. 
+// 'Mono' version of the routine. Returns the number of samples returned in 
+// the "dest" buffer
+uint RateTransposerFloat::transposeStereo(SAMPLETYPE *dest, const SAMPLETYPE *src, uint nSamples)
+{
+    unsigned int srcPos, i, used;
+
+    if (nSamples == 0) return 0;  // no samples, no work
+
+    used = 0;    
+    i = 0;
+
+    // Process the last sample saved from the sPrevSampleLious call first...
+    while (fSlopeCount <= 1.0f) 
+    {
+        dest[2 * i] = (SAMPLETYPE)((1.0f - fSlopeCount) * sPrevSampleL + fSlopeCount * src[0]);
+        dest[2 * i + 1] = (SAMPLETYPE)((1.0f - fSlopeCount) * sPrevSampleR + fSlopeCount * src[1]);
+        i++;
+        fSlopeCount += fRate;
+    }
+    // now always (iSlopeCount > 1.0f)
+    fSlopeCount -= 1.0f;
+
+    if (nSamples > 1)
+    {
+        while (1)
+        {
+            while (fSlopeCount > 1.0f) 
+            {
+                fSlopeCount -= 1.0f;
+                used ++;
+                if (used >= nSamples - 1) goto end;
+            }
+            srcPos = 2 * used;
+
+            dest[2 * i] = (SAMPLETYPE)((1.0f - fSlopeCount) * src[srcPos] 
+                + fSlopeCount * src[srcPos + 2]);
+            dest[2 * i + 1] = (SAMPLETYPE)((1.0f - fSlopeCount) * src[srcPos + 1] 
+                + fSlopeCount * src[srcPos + 3]);
+
+            i++;
+            fSlopeCount += fRate;
+        }
+    }
+end:
+    // Store the last sample for the next round
+    sPrevSampleL = src[2 * nSamples - 2];
+    sPrevSampleR = src[2 * nSamples - 1];
+
+    return i;
+}

Externals/soundtouch/RateTransposer.h

@@ -0,0 +1,159 @@
+////////////////////////////////////////////////////////////////////////////////
+/// 
+/// Sample rate transposer. Changes sample rate by using linear interpolation 
+/// together with anti-alias filtering (first order interpolation with anti-
+/// alias filtering should be quite adequate for this application).
+///
+/// Use either of the derived classes of 'RateTransposerInteger' or 
+/// 'RateTransposerFloat' for corresponding integer/floating point tranposing
+/// algorithm implementation.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2009-02-21 18:00:14 +0200 (Sat, 21 Feb 2009) $
+// File revision : $Revision: 4 $
+//
+// $Id: RateTransposer.h 63 2009-02-21 16:00:14Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef RateTransposer_H
+#define RateTransposer_H
+
+#include <stddef.h>
+#include "AAFilter.h"
+#include "FIFOSamplePipe.h"
+#include "FIFOSampleBuffer.h"
+
+#include "STTypes.h"
+
+namespace soundtouch
+{
+
+/// A common linear samplerate transposer class.
+///
+/// Note: Use function "RateTransposer::newInstance()" to create a new class 
+/// instance instead of the "new" operator; that function automatically 
+/// chooses a correct implementation depending on if integer or floating 
+/// arithmetics are to be used.
+class RateTransposer : public FIFOProcessor
+{
+protected:
+    /// Anti-alias filter object
+    AAFilter *pAAFilter;
+
+    float fRate;
+
+    int numChannels;
+
+    /// Buffer for collecting samples to feed the anti-alias filter between
+    /// two batches
+    FIFOSampleBuffer storeBuffer;
+
+    /// Buffer for keeping samples between transposing & anti-alias filter
+    FIFOSampleBuffer tempBuffer;
+
+    /// Output sample buffer
+    FIFOSampleBuffer outputBuffer;
+
+    BOOL bUseAAFilter;
+
+    virtual void resetRegisters() = 0;
+
+    virtual uint transposeStereo(SAMPLETYPE *dest, 
+                         const SAMPLETYPE *src, 
+                         uint numSamples) = 0;
+    virtual uint transposeMono(SAMPLETYPE *dest, 
+                       const SAMPLETYPE *src, 
+                       uint numSamples) = 0;
+    inline uint transpose(SAMPLETYPE *dest, 
+                   const SAMPLETYPE *src, 
+                   uint numSamples);
+
+    void downsample(const SAMPLETYPE *src, 
+                    uint numSamples);
+    void upsample(const SAMPLETYPE *src, 
+                 uint numSamples);
+
+    /// Transposes sample rate by applying anti-alias filter to prevent folding. 
+    /// Returns amount of samples returned in the "dest" buffer.
+    /// The maximum amount of samples that can be returned at a time is set by
+    /// the 'set_returnBuffer_size' function.
+    void processSamples(const SAMPLETYPE *src, 
+                        uint numSamples);
+
+
+public:
+    RateTransposer();
+    virtual ~RateTransposer();
+
+    /// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+    /// depending on if we're to use integer or floating point arithmetics.
+    static void *operator new(size_t s);
+
+    /// Use this function instead of "new" operator to create a new instance of this class. 
+    /// This function automatically chooses a correct implementation, depending on if 
+    /// integer ot floating point arithmetics are to be used.
+    static RateTransposer *newInstance();
+
+    /// Returns the output buffer object
+    FIFOSamplePipe *getOutput() { return &outputBuffer; };
+
+    /// Returns the store buffer object
+    FIFOSamplePipe *getStore() { return &storeBuffer; };
+
+    /// Return anti-alias filter object
+    AAFilter *getAAFilter();
+
+    /// Enables/disables the anti-alias filter. Zero to disable, nonzero to enable
+    void enableAAFilter(BOOL newMode);
+
+    /// Returns nonzero if anti-alias filter is enabled.
+    BOOL isAAFilterEnabled() const;
+
+    /// Sets new target rate. Normal rate = 1.0, smaller values represent slower 
+    /// rate, larger faster rates.
+    virtual void setRate(float newRate);
+
+    /// Sets the number of channels, 1 = mono, 2 = stereo
+    void setChannels(int channels);
+
+    /// Adds 'numSamples' pcs of samples from the 'samples' memory position into
+    /// the input of the object.
+    void putSamples(const SAMPLETYPE *samples, uint numSamples);
+
+    /// Clears all the samples in the object
+    void clear();
+
+    /// Returns nonzero if there aren't any samples available for outputting.
+    int isEmpty() const;
+};
+
+}
+
+#endif

Externals/soundtouch/STTypes.h

@@ -0,0 +1,191 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Common type definitions for SoundTouch audio processing library.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-12-28 16:53:56 +0200 (Fri, 28 Dec 2012) $
+// File revision : $Revision: 3 $
+//
+// $Id: STTypes.h 162 2012-12-28 14:53:56Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef STTypes_H
+#define STTypes_H
+
+typedef unsigned int    uint;
+typedef unsigned long   ulong;
+
+// Patch for MinGW: on Win64 long is 32-bit
+#ifdef _WIN64
+    typedef unsigned long long ulongptr;
+#else
+    typedef ulong ulongptr;
+#endif
+
+
+// Helper macro for aligning pointer up to next 16-byte boundary
+#define SOUNDTOUCH_ALIGN_POINTER_16(x)      ( ( (ulongptr)(x) + 15 ) & ~(ulongptr)15 )
+
+
+#if (defined(__GNUC__) && !defined(ANDROID))
+    // In GCC, include soundtouch_config.h made by config scritps.
+    // Skip this in Android compilation that uses GCC but without configure scripts.
+    //#include "soundtouch_config.h"
+#endif
+
+#ifndef _WINDEF_
+    // if these aren't defined already by Windows headers, define now
+#if defined(__APPLE__)
+	typedef signed char BOOL;
+#else
+    typedef int BOOL;
+#endif
+    #define FALSE   0
+    #define TRUE    1
+
+#endif  // _WINDEF_
+
+
+namespace soundtouch
+{
+    /// Activate these undef's to overrule the possible sampletype 
+    /// setting inherited from some other header file:
+    #undef SOUNDTOUCH_INTEGER_SAMPLES
+    #undef SOUNDTOUCH_FLOAT_SAMPLES
+
+    #if (defined(__SOFTFP__))
+        // For Android compilation: Force use of Integer samples in case that
+        // compilation uses soft-floating point emulation - soft-fp is way too slow
+        #undef  SOUNDTOUCH_FLOAT_SAMPLES
+        #define SOUNDTOUCH_INTEGER_SAMPLES      1
+    #endif
+
+    #if !(SOUNDTOUCH_INTEGER_SAMPLES || SOUNDTOUCH_FLOAT_SAMPLES)
+       
+        /// Choose either 32bit floating point or 16bit integer sampletype
+        /// by choosing one of the following defines, unless this selection 
+        /// has already been done in some other file.
+        ////
+        /// Notes:
+        /// - In Windows environment, choose the sample format with the
+        ///   following defines.
+        /// - In GNU environment, the floating point samples are used by 
+        ///   default, but integer samples can be chosen by giving the 
+        ///   following switch to the configure script:
+        ///       ./configure --enable-integer-samples
+        ///   However, if you still prefer to select the sample format here 
+        ///   also in GNU environment, then please #undef the INTEGER_SAMPLE
+        ///   and FLOAT_SAMPLE defines first as in comments above.
+        #define SOUNDTOUCH_INTEGER_SAMPLES     1    //< 16bit integer samples
+        //#define SOUNDTOUCH_FLOAT_SAMPLES       1    //< 32bit float samples
+     
+    #endif
+
+    #if (_M_IX86 || __i386__ || __x86_64__ || _M_X64)
+        /// Define this to allow X86-specific assembler/intrinsic optimizations. 
+        /// Notice that library contains also usual C++ versions of each of these
+        /// these routines, so if you're having difficulties getting the optimized 
+        /// routines compiled for whatever reason, you may disable these optimizations 
+        /// to make the library compile.
+
+        #define SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS     1
+
+        /// In GNU environment, allow the user to override this setting by
+        /// giving the following switch to the configure script:
+        /// ./configure --disable-x86-optimizations
+        /// ./configure --enable-x86-optimizations=no
+        #ifdef SOUNDTOUCH_DISABLE_X86_OPTIMIZATIONS
+            #undef SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS
+        #endif
+    #else
+        /// Always disable optimizations when not using a x86 systems.
+        #undef SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS
+
+    #endif
+
+    // If defined, allows the SIMD-optimized routines to take minor shortcuts 
+    // for improved performance. Undefine to require faithfully similar SIMD 
+    // calculations as in normal C implementation.
+    #define SOUNDTOUCH_ALLOW_NONEXACT_SIMD_OPTIMIZATION    1
+
+
+    #ifdef SOUNDTOUCH_INTEGER_SAMPLES
+        // 16bit integer sample type
+        typedef short SAMPLETYPE;
+        // data type for sample accumulation: Use 32bit integer to prevent overflows
+        typedef long  LONG_SAMPLETYPE;
+
+        #ifdef SOUNDTOUCH_FLOAT_SAMPLES
+            // check that only one sample type is defined
+            #error "conflicting sample types defined"
+        #endif // SOUNDTOUCH_FLOAT_SAMPLES
+
+        #ifdef SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS
+            // Allow MMX optimizations
+#ifndef _M_X64
+            #define SOUNDTOUCH_ALLOW_MMX   1
+#endif
+        #endif
+
+    #else
+
+        // floating point samples
+        typedef float  SAMPLETYPE;
+        // data type for sample accumulation: Use double to utilize full precision.
+        typedef double LONG_SAMPLETYPE;
+
+        #ifdef SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS
+            // Allow SSE optimizations
+            #define SOUNDTOUCH_ALLOW_SSE       1
+        #endif
+
+    #endif  // SOUNDTOUCH_INTEGER_SAMPLES
+
+};
+
+// define ST_NO_EXCEPTION_HANDLING switch to disable throwing std exceptions:
+#define ST_NO_EXCEPTION_HANDLING    1
+#ifdef ST_NO_EXCEPTION_HANDLING
+    // Exceptions disabled. Throw asserts instead if enabled.
+    #include <assert.h>
+    #define ST_THROW_RT_ERROR(x)    {assert((const char *)x);}
+#else
+    // use c++ standard exceptions
+    #include <stdexcept>
+    #define ST_THROW_RT_ERROR(x)    {throw std::runtime_error(x);}
+#endif
+
+// When this #define is active, eliminates a clicking sound when the "rate" or "pitch" 
+// parameter setting crosses from value <1 to >=1 or vice versa during processing. 
+// Default is off as such crossover is untypical case and involves a slight sound 
+// quality compromise.
+//#define SOUNDTOUCH_PREVENT_CLICK_AT_RATE_CROSSOVER   1
+
+#endif

Externals/soundtouch/SoundTouch.cpp

@@ -0,0 +1,501 @@
+//////////////////////////////////////////////////////////////////////////////
+///
+/// SoundTouch - main class for tempo/pitch/rate adjusting routines. 
+///
+/// Notes:
+/// - Initialize the SoundTouch object instance by setting up the sound stream 
+///   parameters with functions 'setSampleRate' and 'setChannels', then set 
+///   desired tempo/pitch/rate settings with the corresponding functions.
+///
+/// - The SoundTouch class behaves like a first-in-first-out pipeline: The 
+///   samples that are to be processed are fed into one of the pipe by calling
+///   function 'putSamples', while the ready processed samples can be read 
+///   from the other end of the pipeline with function 'receiveSamples'.
+/// 
+/// - The SoundTouch processing classes require certain sized 'batches' of 
+///   samples in order to process the sound. For this reason the classes buffer 
+///   incoming samples until there are enough of samples available for 
+///   processing, then they carry out the processing step and consequently
+///   make the processed samples available for outputting.
+/// 
+/// - For the above reason, the processing routines introduce a certain 
+///   'latency' between the input and output, so that the samples input to
+///   SoundTouch may not be immediately available in the output, and neither 
+///   the amount of outputtable samples may not immediately be in direct 
+///   relationship with the amount of previously input samples.
+///
+/// - The tempo/pitch/rate control parameters can be altered during processing.
+///   Please notice though that they aren't currently protected by semaphores,
+///   so in multi-thread application external semaphore protection may be
+///   required.
+///
+/// - This class utilizes classes 'TDStretch' for tempo change (without modifying
+///   pitch) and 'RateTransposer' for changing the playback rate (that is, both 
+///   tempo and pitch in the same ratio) of the sound. The third available control 
+///   'pitch' (change pitch but maintain tempo) is produced by a combination of
+///   combining the two other controls.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-06-13 22:29:53 +0300 (Wed, 13 Jun 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: SoundTouch.cpp 143 2012-06-13 19:29:53Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <assert.h>
+#include <stdlib.h>
+#include <memory.h>
+#include <math.h>
+#include <stdio.h>
+
+#include "SoundTouch.h"
+#include "TDStretch.h"
+#include "RateTransposer.h"
+#include "cpu_detect.h"
+
+using namespace soundtouch;
+    
+/// test if two floating point numbers are equal
+#define TEST_FLOAT_EQUAL(a, b)  (fabs(a - b) < 1e-10)
+
+
+/// Print library version string for autoconf
+extern "C" void soundtouch_ac_test()
+{
+    printf("SoundTouch Version: %s\n",SOUNDTOUCH_VERSION);
+} 
+
+
+SoundTouch::SoundTouch()
+{
+    // Initialize rate transposer and tempo changer instances
+
+    pRateTransposer = RateTransposer::newInstance();
+    pTDStretch = TDStretch::newInstance();
+
+    setOutPipe(pTDStretch);
+
+    rate = tempo = 0;
+
+    virtualPitch = 
+    virtualRate = 
+    virtualTempo = 1.0;
+
+    calcEffectiveRateAndTempo();
+
+    channels = 0;
+    bSrateSet = FALSE;
+}
+
+
+
+SoundTouch::~SoundTouch()
+{
+    delete pRateTransposer;
+    delete pTDStretch;
+}
+
+
+
+/// Get SoundTouch library version string
+const char *SoundTouch::getVersionString()
+{
+    static const char *_version = SOUNDTOUCH_VERSION;
+
+    return _version;
+}
+
+
+/// Get SoundTouch library version Id
+uint SoundTouch::getVersionId()
+{
+    return SOUNDTOUCH_VERSION_ID;
+}
+
+
+// Sets the number of channels, 1 = mono, 2 = stereo
+void SoundTouch::setChannels(uint numChannels)
+{
+    if (numChannels != 1 && numChannels != 2) 
+    {
+        ST_THROW_RT_ERROR("Illegal number of channels");
+    }
+    channels = numChannels;
+    pRateTransposer->setChannels((int)numChannels);
+    pTDStretch->setChannels((int)numChannels);
+}
+
+
+
+// Sets new rate control value. Normal rate = 1.0, smaller values
+// represent slower rate, larger faster rates.
+void SoundTouch::setRate(float newRate)
+{
+    virtualRate = newRate;
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets new rate control value as a difference in percents compared
+// to the original rate (-50 .. +100 %)
+void SoundTouch::setRateChange(float newRate)
+{
+    virtualRate = 1.0f + 0.01f * newRate;
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets new tempo control value. Normal tempo = 1.0, smaller values
+// represent slower tempo, larger faster tempo.
+void SoundTouch::setTempo(float newTempo)
+{
+    virtualTempo = newTempo;
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets new tempo control value as a difference in percents compared
+// to the original tempo (-50 .. +100 %)
+void SoundTouch::setTempoChange(float newTempo)
+{
+    virtualTempo = 1.0f + 0.01f * newTempo;
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets new pitch control value. Original pitch = 1.0, smaller values
+// represent lower pitches, larger values higher pitch.
+void SoundTouch::setPitch(float newPitch)
+{
+    virtualPitch = newPitch;
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets pitch change in octaves compared to the original pitch
+// (-1.00 .. +1.00)
+void SoundTouch::setPitchOctaves(float newPitch)
+{
+    virtualPitch = (float)exp(0.69314718056f * newPitch);
+    calcEffectiveRateAndTempo();
+}
+
+
+
+// Sets pitch change in semi-tones compared to the original pitch
+// (-12 .. +12)
+void SoundTouch::setPitchSemiTones(int newPitch)
+{
+    setPitchOctaves((float)newPitch / 12.0f);
+}
+
+
+
+void SoundTouch::setPitchSemiTones(float newPitch)
+{
+    setPitchOctaves(newPitch / 12.0f);
+}
+
+
+// Calculates 'effective' rate and tempo values from the
+// nominal control values.
+void SoundTouch::calcEffectiveRateAndTempo()
+{
+    float oldTempo = tempo;
+    float oldRate = rate;
+
+    tempo = virtualTempo / virtualPitch;
+    rate = virtualPitch * virtualRate;
+
+    if (!TEST_FLOAT_EQUAL(rate,oldRate)) pRateTransposer->setRate(rate);
+    if (!TEST_FLOAT_EQUAL(tempo, oldTempo)) pTDStretch->setTempo(tempo);
+
+#ifndef SOUNDTOUCH_PREVENT_CLICK_AT_RATE_CROSSOVER
+    if (rate <= 1.0f) 
+    {
+        if (output != pTDStretch) 
+        {
+            FIFOSamplePipe *tempoOut;
+
+            assert(output == pRateTransposer);
+            // move samples in the current output buffer to the output of pTDStretch
+            tempoOut = pTDStretch->getOutput();
+            tempoOut->moveSamples(*output);
+            // move samples in pitch transposer's store buffer to tempo changer's input
+            pTDStretch->moveSamples(*pRateTransposer->getStore());
+
+            output = pTDStretch;
+        }
+    }
+    else
+#endif
+    {
+        if (output != pRateTransposer) 
+        {
+            FIFOSamplePipe *transOut;
+
+            assert(output == pTDStretch);
+            // move samples in the current output buffer to the output of pRateTransposer
+            transOut = pRateTransposer->getOutput();
+            transOut->moveSamples(*output);
+            // move samples in tempo changer's input to pitch transposer's input
+            pRateTransposer->moveSamples(*pTDStretch->getInput());
+
+            output = pRateTransposer;
+        }
+    } 
+}
+
+
+// Sets sample rate.
+void SoundTouch::setSampleRate(uint srate)
+{
+    bSrateSet = TRUE;
+    // set sample rate, leave other tempo changer parameters as they are.
+    pTDStretch->setParameters((int)srate);
+}
+
+
+// Adds 'numSamples' pcs of samples from the 'samples' memory position into
+// the input of the object.
+void SoundTouch::putSamples(const SAMPLETYPE *samples, uint nSamples)
+{
+    if (bSrateSet == FALSE) 
+    {
+        ST_THROW_RT_ERROR("SoundTouch : Sample rate not defined");
+    } 
+    else if (channels == 0) 
+    {
+        ST_THROW_RT_ERROR("SoundTouch : Number of channels not defined");
+    }
+
+    // Transpose the rate of the new samples if necessary
+    /* Bypass the nominal setting - can introduce a click in sound when tempo/pitch control crosses the nominal value...
+    if (rate == 1.0f) 
+    {
+        // The rate value is same as the original, simply evaluate the tempo changer. 
+        assert(output == pTDStretch);
+        if (pRateTransposer->isEmpty() == 0) 
+        {
+            // yet flush the last samples in the pitch transposer buffer
+            // (may happen if 'rate' changes from a non-zero value to zero)
+            pTDStretch->moveSamples(*pRateTransposer);
+        }
+        pTDStretch->putSamples(samples, nSamples);
+    } 
+    */
+#ifndef SOUNDTOUCH_PREVENT_CLICK_AT_RATE_CROSSOVER
+    else if (rate <= 1.0f) 
+    {
+        // transpose the rate down, output the transposed sound to tempo changer buffer
+        assert(output == pTDStretch);
+        pRateTransposer->putSamples(samples, nSamples);
+        pTDStretch->moveSamples(*pRateTransposer);
+    } 
+    else 
+#endif
+    {
+        // evaluate the tempo changer, then transpose the rate up, 
+        assert(output == pRateTransposer);
+        pTDStretch->putSamples(samples, nSamples);
+        pRateTransposer->moveSamples(*pTDStretch);
+    }
+}
+
+
+// Flushes the last samples from the processing pipeline to the output.
+// Clears also the internal processing buffers.
+//
+// Note: This function is meant for extracting the last samples of a sound
+// stream. This function may introduce additional blank samples in the end
+// of the sound stream, and thus it's not recommended to call this function
+// in the middle of a sound stream.
+void SoundTouch::flush()
+{
+    int i;
+    int nUnprocessed;
+    int nOut;
+    SAMPLETYPE buff[64*2];   // note: allocate 2*64 to cater 64 sample frames of stereo sound
+
+    // check how many samples still await processing, and scale
+    // that by tempo & rate to get expected output sample count
+    nUnprocessed = numUnprocessedSamples();
+    nUnprocessed = (int)((double)nUnprocessed / (tempo * rate) + 0.5);
+
+    nOut = numSamples();        // ready samples currently in buffer ...
+    nOut += nUnprocessed;       // ... and how many we expect there to be in the end
+    
+    memset(buff, 0, 64 * channels * sizeof(SAMPLETYPE));
+    // "Push" the last active samples out from the processing pipeline by
+    // feeding blank samples into the processing pipeline until new, 
+    // processed samples appear in the output (not however, more than 
+    // 8ksamples in any case)
+    for (i = 0; i < 128; i ++) 
+    {
+        putSamples(buff, 64);
+        if ((int)numSamples() >= nOut) 
+        {
+            // Enough new samples have appeared into the output!
+            // As samples come from processing with bigger chunks, now truncate it
+            // back to maximum "nOut" samples to improve duration accuracy 
+            adjustAmountOfSamples(nOut);
+
+            // finish
+            break;  
+        }
+    }
+
+    // Clear working buffers
+    pRateTransposer->clear();
+    pTDStretch->clearInput();
+    // yet leave the 'tempoChanger' output intouched as that's where the
+    // flushed samples are!
+}
+
+
+// Changes a setting controlling the processing system behaviour. See the
+// 'SETTING_...' defines for available setting ID's.
+BOOL SoundTouch::setSetting(int settingId, int value)
+{
+    int sampleRate, sequenceMs, seekWindowMs, overlapMs;
+
+    // read current tdstretch routine parameters
+    pTDStretch->getParameters(&sampleRate, &sequenceMs, &seekWindowMs, &overlapMs);
+
+    switch (settingId) 
+    {
+        case SETTING_USE_AA_FILTER :
+            // enables / disabless anti-alias filter
+            pRateTransposer->enableAAFilter((value != 0) ? TRUE : FALSE);
+            return TRUE;
+
+        case SETTING_AA_FILTER_LENGTH :
+            // sets anti-alias filter length
+            pRateTransposer->getAAFilter()->setLength(value);
+            return TRUE;
+
+        case SETTING_USE_QUICKSEEK :
+            // enables / disables tempo routine quick seeking algorithm
+            pTDStretch->enableQuickSeek((value != 0) ? TRUE : FALSE);
+            return TRUE;
+
+        case SETTING_SEQUENCE_MS:
+            // change time-stretch sequence duration parameter
+            pTDStretch->setParameters(sampleRate, value, seekWindowMs, overlapMs);
+            return TRUE;
+
+        case SETTING_SEEKWINDOW_MS:
+            // change time-stretch seek window length parameter
+            pTDStretch->setParameters(sampleRate, sequenceMs, value, overlapMs);
+            return TRUE;
+
+        case SETTING_OVERLAP_MS:
+            // change time-stretch overlap length parameter
+            pTDStretch->setParameters(sampleRate, sequenceMs, seekWindowMs, value);
+            return TRUE;
+
+        default :
+            return FALSE;
+    }
+}
+
+
+// Reads a setting controlling the processing system behaviour. See the
+// 'SETTING_...' defines for available setting ID's.
+//
+// Returns the setting value.
+int SoundTouch::getSetting(int settingId) const
+{
+    int temp;
+
+    switch (settingId) 
+    {
+        case SETTING_USE_AA_FILTER :
+            return (uint)pRateTransposer->isAAFilterEnabled();
+
+        case SETTING_AA_FILTER_LENGTH :
+            return pRateTransposer->getAAFilter()->getLength();
+
+        case SETTING_USE_QUICKSEEK :
+            return (uint)   pTDStretch->isQuickSeekEnabled();
+
+        case SETTING_SEQUENCE_MS:
+            pTDStretch->getParameters(NULL, &temp, NULL, NULL);
+            return temp;
+
+        case SETTING_SEEKWINDOW_MS:
+            pTDStretch->getParameters(NULL, NULL, &temp, NULL);
+            return temp;
+
+        case SETTING_OVERLAP_MS:
+            pTDStretch->getParameters(NULL, NULL, NULL, &temp);
+            return temp;
+
+		case SETTING_NOMINAL_INPUT_SEQUENCE :
+			return pTDStretch->getInputSampleReq();
+
+		case SETTING_NOMINAL_OUTPUT_SEQUENCE :
+			return pTDStretch->getOutputBatchSize();
+
+		default :
+            return 0;
+    }
+}
+
+
+// Clears all the samples in the object's output and internal processing
+// buffers.
+void SoundTouch::clear()
+{
+    pRateTransposer->clear();
+    pTDStretch->clear();
+}
+
+
+
+/// Returns number of samples currently unprocessed.
+uint SoundTouch::numUnprocessedSamples() const
+{
+    FIFOSamplePipe * psp;
+    if (pTDStretch)
+    {
+        psp = pTDStretch->getInput();
+        if (psp)
+        {
+            return psp->numSamples();
+        }
+    }
+    return 0;
+}

Externals/soundtouch/SoundTouch.h

@@ -0,0 +1,277 @@
+//////////////////////////////////////////////////////////////////////////////
+///
+/// SoundTouch - main class for tempo/pitch/rate adjusting routines. 
+///
+/// Notes:
+/// - Initialize the SoundTouch object instance by setting up the sound stream 
+///   parameters with functions 'setSampleRate' and 'setChannels', then set 
+///   desired tempo/pitch/rate settings with the corresponding functions.
+///
+/// - The SoundTouch class behaves like a first-in-first-out pipeline: The 
+///   samples that are to be processed are fed into one of the pipe by calling
+///   function 'putSamples', while the ready processed samples can be read 
+///   from the other end of the pipeline with function 'receiveSamples'.
+/// 
+/// - The SoundTouch processing classes require certain sized 'batches' of 
+///   samples in order to process the sound. For this reason the classes buffer 
+///   incoming samples until there are enough of samples available for 
+///   processing, then they carry out the processing step and consequently
+///   make the processed samples available for outputting.
+/// 
+/// - For the above reason, the processing routines introduce a certain 
+///   'latency' between the input and output, so that the samples input to
+///   SoundTouch may not be immediately available in the output, and neither 
+///   the amount of outputtable samples may not immediately be in direct 
+///   relationship with the amount of previously input samples.
+///
+/// - The tempo/pitch/rate control parameters can be altered during processing.
+///   Please notice though that they aren't currently protected by semaphores,
+///   so in multi-thread application external semaphore protection may be
+///   required.
+///
+/// - This class utilizes classes 'TDStretch' for tempo change (without modifying
+///   pitch) and 'RateTransposer' for changing the playback rate (that is, both 
+///   tempo and pitch in the same ratio) of the sound. The third available control 
+///   'pitch' (change pitch but maintain tempo) is produced by a combination of
+///   combining the two other controls.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-12-28 21:32:59 +0200 (Fri, 28 Dec 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: SoundTouch.h 163 2012-12-28 19:32:59Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef SoundTouch_H
+#define SoundTouch_H
+
+#include "FIFOSamplePipe.h"
+#include "STTypes.h"
+
+namespace soundtouch
+{
+
+/// Soundtouch library version string
+#define SOUNDTOUCH_VERSION          "1.7.1"
+
+/// SoundTouch library version id
+#define SOUNDTOUCH_VERSION_ID       (10701)
+
+//
+// Available setting IDs for the 'setSetting' & 'get_setting' functions:
+
+/// Enable/disable anti-alias filter in pitch transposer (0 = disable)
+#define SETTING_USE_AA_FILTER       0
+
+/// Pitch transposer anti-alias filter length (8 .. 128 taps, default = 32)
+#define SETTING_AA_FILTER_LENGTH    1
+
+/// Enable/disable quick seeking algorithm in tempo changer routine
+/// (enabling quick seeking lowers CPU utilization but causes a minor sound
+///  quality compromising)
+#define SETTING_USE_QUICKSEEK       2
+
+/// Time-stretch algorithm single processing sequence length in milliseconds. This determines 
+/// to how long sequences the original sound is chopped in the time-stretch algorithm. 
+/// See "STTypes.h" or README for more information.
+#define SETTING_SEQUENCE_MS         3
+
+/// Time-stretch algorithm seeking window length in milliseconds for algorithm that finds the 
+/// best possible overlapping location. This determines from how wide window the algorithm 
+/// may look for an optimal joining location when mixing the sound sequences back together. 
+/// See "STTypes.h" or README for more information.
+#define SETTING_SEEKWINDOW_MS       4
+
+/// Time-stretch algorithm overlap length in milliseconds. When the chopped sound sequences 
+/// are mixed back together, to form a continuous sound stream, this parameter defines over 
+/// how long period the two consecutive sequences are let to overlap each other. 
+/// See "STTypes.h" or README for more information.
+#define SETTING_OVERLAP_MS          5
+
+
+/// Call "getSetting" with this ID to query nominal average processing sequence
+/// size in samples. This value tells approcimate value how many input samples 
+/// SoundTouch needs to gather before it does DSP processing run for the sample batch.
+///
+/// Notices: 
+/// - This is read-only parameter, i.e. setSetting ignores this parameter
+/// - Returned value is approximate average value, exact processing batch
+///   size may wary from time to time
+/// - This parameter value is not constant but may change depending on 
+///   tempo/pitch/rate/samplerate settings.
+#define SETTING_NOMINAL_INPUT_SEQUENCE		6
+
+
+/// Call "getSetting" with this ID to query nominal average processing output 
+/// size in samples. This value tells approcimate value how many output samples 
+/// SoundTouch outputs once it does DSP processing run for a batch of input samples.
+///	
+/// Notices: 
+/// - This is read-only parameter, i.e. setSetting ignores this parameter
+/// - Returned value is approximate average value, exact processing batch
+///   size may wary from time to time
+/// - This parameter value is not constant but may change depending on 
+///   tempo/pitch/rate/samplerate settings.
+#define SETTING_NOMINAL_OUTPUT_SEQUENCE		7
+
+class SoundTouch : public FIFOProcessor
+{
+private:
+    /// Rate transposer class instance
+    class RateTransposer *pRateTransposer;
+
+    /// Time-stretch class instance
+    class TDStretch *pTDStretch;
+
+    /// Virtual pitch parameter. Effective rate & tempo are calculated from these parameters.
+    float virtualRate;
+
+    /// Virtual pitch parameter. Effective rate & tempo are calculated from these parameters.
+    float virtualTempo;
+
+    /// Virtual pitch parameter. Effective rate & tempo are calculated from these parameters.
+    float virtualPitch;
+
+    /// Flag: Has sample rate been set?
+    BOOL  bSrateSet;
+
+    /// Calculates effective rate & tempo valuescfrom 'virtualRate', 'virtualTempo' and 
+    /// 'virtualPitch' parameters.
+    void calcEffectiveRateAndTempo();
+
+protected :
+    /// Number of channels
+    uint  channels;
+
+    /// Effective 'rate' value calculated from 'virtualRate', 'virtualTempo' and 'virtualPitch'
+    float rate;
+
+    /// Effective 'tempo' value calculated from 'virtualRate', 'virtualTempo' and 'virtualPitch'
+    float tempo;
+
+public:
+    SoundTouch();
+    virtual ~SoundTouch();
+
+    /// Get SoundTouch library version string
+    static const char *getVersionString();
+
+    /// Get SoundTouch library version Id
+    static uint getVersionId();
+
+    /// Sets new rate control value. Normal rate = 1.0, smaller values
+    /// represent slower rate, larger faster rates.
+    void setRate(float newRate);
+
+    /// Sets new tempo control value. Normal tempo = 1.0, smaller values
+    /// represent slower tempo, larger faster tempo.
+    void setTempo(float newTempo);
+
+    /// Sets new rate control value as a difference in percents compared
+    /// to the original rate (-50 .. +100 %)
+    void setRateChange(float newRate);
+
+    /// Sets new tempo control value as a difference in percents compared
+    /// to the original tempo (-50 .. +100 %)
+    void setTempoChange(float newTempo);
+
+    /// Sets new pitch control value. Original pitch = 1.0, smaller values
+    /// represent lower pitches, larger values higher pitch.
+    void setPitch(float newPitch);
+
+    /// Sets pitch change in octaves compared to the original pitch  
+    /// (-1.00 .. +1.00)
+    void setPitchOctaves(float newPitch);
+
+    /// Sets pitch change in semi-tones compared to the original pitch
+    /// (-12 .. +12)
+    void setPitchSemiTones(int newPitch);
+    void setPitchSemiTones(float newPitch);
+
+    /// Sets the number of channels, 1 = mono, 2 = stereo
+    void setChannels(uint numChannels);
+
+    /// Sets sample rate.
+    void setSampleRate(uint srate);
+
+    /// Flushes the last samples from the processing pipeline to the output.
+    /// Clears also the internal processing buffers.
+    //
+    /// Note: This function is meant for extracting the last samples of a sound
+    /// stream. This function may introduce additional blank samples in the end
+    /// of the sound stream, and thus it's not recommended to call this function
+    /// in the middle of a sound stream.
+    void flush();
+
+    /// Adds 'numSamples' pcs of samples from the 'samples' memory position into
+    /// the input of the object. Notice that sample rate _has_to_ be set before
+    /// calling this function, otherwise throws a runtime_error exception.
+    virtual void putSamples(
+            const SAMPLETYPE *samples,  ///< Pointer to sample buffer.
+            uint numSamples                         ///< Number of samples in buffer. Notice
+                                                    ///< that in case of stereo-sound a single sample
+                                                    ///< contains data for both channels.
+            );
+
+    /// Clears all the samples in the object's output and internal processing
+    /// buffers.
+    virtual void clear();
+
+    /// Changes a setting controlling the processing system behaviour. See the
+    /// 'SETTING_...' defines for available setting ID's.
+    /// 
+    /// \return 'TRUE' if the setting was succesfully changed
+    BOOL setSetting(int settingId,   ///< Setting ID number. see SETTING_... defines.
+                    int value        ///< New setting value.
+                    );
+
+    /// Reads a setting controlling the processing system behaviour. See the
+    /// 'SETTING_...' defines for available setting ID's.
+    ///
+    /// \return the setting value.
+    int getSetting(int settingId    ///< Setting ID number, see SETTING_... defines.
+                   ) const;
+
+    /// Returns number of samples currently unprocessed.
+    virtual uint numUnprocessedSamples() const;
+
+
+    /// Other handy functions that are implemented in the ancestor classes (see
+    /// classes 'FIFOProcessor' and 'FIFOSamplePipe')
+    ///
+    /// - receiveSamples() : Use this function to receive 'ready' processed samples from SoundTouch.
+    /// - numSamples()     : Get number of 'ready' samples that can be received with 
+    ///                      function 'receiveSamples()'
+    /// - isEmpty()        : Returns nonzero if there aren't any 'ready' samples.
+    /// - clear()          : Clears all samples from ready/processing buffers.
+};
+
+}
+#endif

Externals/soundtouch/SoundTouch.vcxproj

@@ -0,0 +1,352 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{68A5DD20-7057-448B-8FE0-B6AC8D205509}</ProjectGuid>
+    <SccProjectName />
+    <SccLocalPath />
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseOfMfc>false</UseOfMfc>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseOfMfc>false</UseOfMfc>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseOfMfc>false</UseOfMfc>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseOfMfc>false</UseOfMfc>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup>
+    <_ProjectFileVersion>10.0.40219.1</_ProjectFileVersion>
+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">Release\</OutDir>
+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Release|x64'">Release\</OutDir>
+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">Release\</IntDir>
+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Release|x64'">Release\</IntDir>
+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Debug\</OutDir>
+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Debug\</OutDir>
+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Debug\</IntDir>
+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Debug\</IntDir>
+    <CodeAnalysisRuleSet Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">AllRules.ruleset</CodeAnalysisRuleSet>
+    <CodeAnalysisRuleSet Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">AllRules.ruleset</CodeAnalysisRuleSet>
+    <CodeAnalysisRules Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" />
+    <CodeAnalysisRules Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" />
+    <CodeAnalysisRuleAssemblies Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" />
+    <CodeAnalysisRuleAssemblies Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" />
+    <CodeAnalysisRuleSet Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">AllRules.ruleset</CodeAnalysisRuleSet>
+    <CodeAnalysisRuleSet Condition="'$(Configuration)|$(Platform)'=='Release|x64'">AllRules.ruleset</CodeAnalysisRuleSet>
+    <CodeAnalysisRules Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" />
+    <CodeAnalysisRules Condition="'$(Configuration)|$(Platform)'=='Release|x64'" />
+    <CodeAnalysisRuleAssemblies Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" />
+    <CodeAnalysisRuleAssemblies Condition="'$(Configuration)|$(Platform)'=='Release|x64'" />
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <Optimization>Full</Optimization>
+      <InlineFunctionExpansion>AnySuitable</InlineFunctionExpansion>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <StringPooling>true</StringPooling>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <PrecompiledHeaderOutputFile>Release/SoundTouch.pch</PrecompiledHeaderOutputFile>
+      <AssemblerListingLocation>Release/</AssemblerListingLocation>
+      <ObjectFileName>Release/</ObjectFileName>
+      <ProgramDataBaseFileName>Release/</ProgramDataBaseFileName>
+      <WarningLevel>Level3</WarningLevel>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+      <DebugInformationFormat>
+      </DebugInformationFormat>
+      <CompileAs>Default</CompileAs>
+    </ClCompile>
+    <Lib>
+      <OutputFile>Win32\SoundTouch.lib</OutputFile>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+    </Lib>
+    <PostBuildEvent>
+      <Command>
+      </Command>
+    </PostBuildEvent>
+    <ResourceCompile>
+      <PreprocessorDefinitions>NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <Culture>0x040b</Culture>
+    </ResourceCompile>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <Optimization>Full</Optimization>
+      <InlineFunctionExpansion>AnySuitable</InlineFunctionExpansion>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <StringPooling>true</StringPooling>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <PrecompiledHeaderOutputFile>Release/SoundTouch.pch</PrecompiledHeaderOutputFile>
+      <AssemblerListingLocation>Release/</AssemblerListingLocation>
+      <ObjectFileName>Release/</ObjectFileName>
+      <ProgramDataBaseFileName>Release/</ProgramDataBaseFileName>
+      <WarningLevel>Level3</WarningLevel>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+      <DebugInformationFormat>
+      </DebugInformationFormat>
+      <CompileAs>Default</CompileAs>
+      <EnableEnhancedInstructionSet>StreamingSIMDExtensions2</EnableEnhancedInstructionSet>
+    </ClCompile>
+    <Lib>
+      <OutputFile>Win64\SoundTouch.lib</OutputFile>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+    </Lib>
+    <PostBuildEvent>
+      <Command>
+      </Command>
+    </PostBuildEvent>
+    <ResourceCompile>
+      <PreprocessorDefinitions>NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <Culture>0x040b</Culture>
+    </ResourceCompile>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <Optimization>Disabled</Optimization>
+      <AdditionalIncludeDirectories>%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <PrecompiledHeaderOutputFile>Debug/SoundTouch.pch</PrecompiledHeaderOutputFile>
+      <AssemblerListingLocation>Debug/</AssemblerListingLocation>
+      <ObjectFileName>Debug/</ObjectFileName>
+      <ProgramDataBaseFileName>Debug/</ProgramDataBaseFileName>
+      <BrowseInformation>true</BrowseInformation>
+      <WarningLevel>Level3</WarningLevel>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+      <DebugInformationFormat>EditAndContinue</DebugInformationFormat>
+      <CompileAs>Default</CompileAs>
+    </ClCompile>
+    <Lib>
+      <OutputFile>Win32\SoundTouch.lib</OutputFile>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+    </Lib>
+    <PostBuildEvent>
+      <Command>
+      </Command>
+    </PostBuildEvent>
+    <ResourceCompile>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <Culture>0x040b</Culture>
+    </ResourceCompile>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <Optimization>Disabled</Optimization>
+      <AdditionalIncludeDirectories>%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <PrecompiledHeaderOutputFile>Debug/SoundTouch.pch</PrecompiledHeaderOutputFile>
+      <AssemblerListingLocation>Debug/</AssemblerListingLocation>
+      <ObjectFileName>Debug/</ObjectFileName>
+      <ProgramDataBaseFileName>Debug/</ProgramDataBaseFileName>
+      <BrowseInformation>true</BrowseInformation>
+      <WarningLevel>Level3</WarningLevel>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+      <DebugInformationFormat>ProgramDatabase</DebugInformationFormat>
+      <CompileAs>Default</CompileAs>
+    </ClCompile>
+    <Lib>
+      <OutputFile>Win64\SoundTouchD.lib</OutputFile>
+      <SuppressStartupBanner>true</SuppressStartupBanner>
+    </Lib>
+    <PostBuildEvent>
+      <Command>
+      </Command>
+    </PostBuildEvent>
+    <ResourceCompile>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <Culture>0x040b</Culture>
+    </ResourceCompile>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="AAFilter.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="cpu_detect_x86.cpp" />
+    <ClCompile Include="FIFOSampleBuffer.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="FIRFilter.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="mmx_optimized.cpp" />
+    <ClCompile Include="RateTransposer.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="SoundTouch.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="sse_optimized.cpp" />
+    <ClCompile Include="TDStretch.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Disabled</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Disabled</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">EnableFastChecks</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">EnableFastChecks</BasicRuntimeChecks>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</BrowseInformation>
+      <BrowseInformation Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</BrowseInformation>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Release|x64'">MaxSpeed</Optimization>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions Condition="'$(Configuration)|$(Platform)'=='Release|x64'">%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <ClCompile Include="BPMDetect.cpp" />
+    <ClCompile Include="PeakFinder.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="AAFilter.h" />
+    <ClInclude Include="BPMDetect.h" />
+    <ClInclude Include="cpu_detect.h" />
+    <ClInclude Include="FIFOSampleBuffer.h" />
+    <ClInclude Include="FIFOSamplePipe.h" />
+    <ClInclude Include="FIRFilter.h" />
+    <ClInclude Include="PeakFinder.h" />
+    <ClInclude Include="RateTransposer.h" />
+    <ClInclude Include="SoundTouch.h" />
+    <ClInclude Include="STTypes.h" />
+    <ClInclude Include="TDStretch.h" />
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>

Externals/soundtouch/SoundTouch.vcxproj.filters

@@ -0,0 +1,60 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{b7786182-6345-4203-8b48-39eec5ec85dc}</UniqueIdentifier>
+      <Extensions>cpp;c;cxx;rc;def;r;odl;idl;hpj;bat</Extensions>
+    </Filter>
+    <Filter Include="Source Files\bpm">
+      <UniqueIdentifier>{75380bb9-1e58-4186-a9cd-ec7cd284e6a5}</UniqueIdentifier>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="AAFilter.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="cpu_detect_x86.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="FIFOSampleBuffer.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="FIRFilter.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="mmx_optimized.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="RateTransposer.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="SoundTouch.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="sse_optimized.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="TDStretch.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="BPMDetect.cpp">
+      <Filter>Source Files\bpm</Filter>
+    </ClCompile>
+    <ClCompile Include="PeakFinder.cpp">
+      <Filter>Source Files\bpm</Filter>
+    </ClCompile>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="FIFOSampleBuffer.h" />
+    <ClInclude Include="FIFOSamplePipe.h" />
+    <ClInclude Include="FIRFilter.h" />
+    <ClInclude Include="PeakFinder.h" />
+    <ClInclude Include="RateTransposer.h" />
+    <ClInclude Include="SoundTouch.h" />
+    <ClInclude Include="STTypes.h" />
+    <ClInclude Include="TDStretch.h" />
+    <ClInclude Include="AAFilter.h" />
+    <ClInclude Include="BPMDetect.h" />
+    <ClInclude Include="cpu_detect.h" />
+  </ItemGroup>
+</Project>

Externals/soundtouch/TDStretch.cpp

@@ -0,0 +1,808 @@
+////////////////////////////////////////////////////////////////////////////////
+/// 
+/// Sampled sound tempo changer/time stretch algorithm. Changes the sound tempo 
+/// while maintaining the original pitch by using a time domain WSOLA-like 
+/// method with several performance-increasing tweaks.
+///
+/// Note : MMX optimized functions reside in a separate, platform-specific 
+/// file, e.g. 'mmx_win.cpp' or 'mmx_gcc.cpp'
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-11-08 20:53:01 +0200 (Thu, 08 Nov 2012) $
+// File revision : $Revision: 1.12 $
+//
+// $Id: TDStretch.cpp 160 2012-11-08 18:53:01Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include <string.h>
+#include <limits.h>
+#include <assert.h>
+#include <math.h>
+#include <float.h>
+
+#include "STTypes.h"
+#include "cpu_detect.h"
+#include "TDStretch.h"
+
+#include <stdio.h>
+
+using namespace soundtouch;
+
+#define max(x, y) (((x) > (y)) ? (x) : (y))
+
+
+/*****************************************************************************
+ *
+ * Constant definitions
+ *
+ *****************************************************************************/
+
+// Table for the hierarchical mixing position seeking algorithm
+static const short _scanOffsets[5][24]={
+    { 124,  186,  248,  310,  372,  434,  496,  558,  620,  682,  744, 806,
+      868,  930,  992, 1054, 1116, 1178, 1240, 1302, 1364, 1426, 1488,   0},
+    {-100,  -75,  -50,  -25,   25,   50,   75,  100,    0,    0,    0,   0,
+        0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,   0},
+    { -20,  -15,  -10,   -5,    5,   10,   15,   20,    0,    0,    0,   0,
+        0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,   0},
+    {  -4,   -3,   -2,   -1,    1,    2,    3,    4,    0,    0,    0,   0,
+        0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,   0},
+    { 121,  114,   97,  114,   98,  105,  108,   32,  104,   99,  117,  111,
+      116,  100,  110,  117,  111,  115,    0,    0,    0,    0,    0,   0}};
+
+/*****************************************************************************
+ *
+ * Implementation of the class 'TDStretch'
+ *
+ *****************************************************************************/
+
+
+TDStretch::TDStretch() : FIFOProcessor(&outputBuffer)
+{
+    bQuickSeek = FALSE;
+    channels = 2;
+
+    pMidBuffer = NULL;
+    pMidBufferUnaligned = NULL;
+    overlapLength = 0;
+
+    bAutoSeqSetting = TRUE;
+    bAutoSeekSetting = TRUE;
+
+//    outDebt = 0;
+    skipFract = 0;
+
+    tempo = 1.0f;
+    setParameters(44100, DEFAULT_SEQUENCE_MS, DEFAULT_SEEKWINDOW_MS, DEFAULT_OVERLAP_MS);
+    setTempo(1.0f);
+
+    clear();
+}
+
+
+
+TDStretch::~TDStretch()
+{
+    delete[] pMidBufferUnaligned;
+}
+
+
+
+// Sets routine control parameters. These control are certain time constants
+// defining how the sound is stretched to the desired duration.
+//
+// 'sampleRate' = sample rate of the sound
+// 'sequenceMS' = one processing sequence length in milliseconds (default = 82 ms)
+// 'seekwindowMS' = seeking window length for scanning the best overlapping 
+//      position (default = 28 ms)
+// 'overlapMS' = overlapping length (default = 12 ms)
+
+void TDStretch::setParameters(int aSampleRate, int aSequenceMS, 
+                              int aSeekWindowMS, int aOverlapMS)
+{
+    // accept only positive parameter values - if zero or negative, use old values instead
+    if (aSampleRate > 0)   this->sampleRate = aSampleRate;
+    if (aOverlapMS > 0)    this->overlapMs = aOverlapMS;
+
+    if (aSequenceMS > 0)
+    {
+        this->sequenceMs = aSequenceMS;
+        bAutoSeqSetting = FALSE;
+    } 
+    else if (aSequenceMS == 0)
+    {
+        // if zero, use automatic setting
+        bAutoSeqSetting = TRUE;
+    }
+
+    if (aSeekWindowMS > 0) 
+    {
+        this->seekWindowMs = aSeekWindowMS;
+        bAutoSeekSetting = FALSE;
+    } 
+    else if (aSeekWindowMS == 0) 
+    {
+        // if zero, use automatic setting
+        bAutoSeekSetting = TRUE;
+    }
+
+    calcSeqParameters();
+
+    calculateOverlapLength(overlapMs);
+
+    // set tempo to recalculate 'sampleReq'
+    setTempo(tempo);
+
+}
+
+
+
+/// Get routine control parameters, see setParameters() function.
+/// Any of the parameters to this function can be NULL, in such case corresponding parameter
+/// value isn't returned.
+void TDStretch::getParameters(int *pSampleRate, int *pSequenceMs, int *pSeekWindowMs, int *pOverlapMs) const
+{
+    if (pSampleRate)
+    {
+        *pSampleRate = sampleRate;
+    }
+
+    if (pSequenceMs)
+    {
+        *pSequenceMs = (bAutoSeqSetting) ? (USE_AUTO_SEQUENCE_LEN) : sequenceMs;
+    }
+
+    if (pSeekWindowMs)
+    {
+        *pSeekWindowMs = (bAutoSeekSetting) ? (USE_AUTO_SEEKWINDOW_LEN) : seekWindowMs;
+    }
+
+    if (pOverlapMs)
+    {
+        *pOverlapMs = overlapMs;
+    }
+}
+
+
+// Overlaps samples in 'midBuffer' with the samples in 'pInput'
+void TDStretch::overlapMono(SAMPLETYPE *pOutput, const SAMPLETYPE *pInput) const
+{
+    int i;
+    SAMPLETYPE m1, m2;
+
+    m1 = (SAMPLETYPE)0;
+    m2 = (SAMPLETYPE)overlapLength;
+
+    for (i = 0; i < overlapLength ; i ++) 
+    {
+        pOutput[i] = (pInput[i] * m1 + pMidBuffer[i] * m2 ) / overlapLength;
+        m1 += 1;
+        m2 -= 1;
+    }
+}
+
+
+
+void TDStretch::clearMidBuffer()
+{
+    memset(pMidBuffer, 0, 2 * sizeof(SAMPLETYPE) * overlapLength);
+}
+
+
+void TDStretch::clearInput()
+{
+    inputBuffer.clear();
+    clearMidBuffer();
+}
+
+
+// Clears the sample buffers
+void TDStretch::clear()
+{
+    outputBuffer.clear();
+    clearInput();
+}
+
+
+
+// Enables/disables the quick position seeking algorithm. Zero to disable, nonzero
+// to enable
+void TDStretch::enableQuickSeek(BOOL enable)
+{
+    bQuickSeek = enable;
+}
+
+
+// Returns nonzero if the quick seeking algorithm is enabled.
+BOOL TDStretch::isQuickSeekEnabled() const
+{
+    return bQuickSeek;
+}
+
+
+// Seeks for the optimal overlap-mixing position.
+int TDStretch::seekBestOverlapPosition(const SAMPLETYPE *refPos)
+{
+    if (bQuickSeek) 
+    {
+        return seekBestOverlapPositionQuick(refPos);
+    } 
+    else 
+    {
+        return seekBestOverlapPositionFull(refPos);
+    }
+}
+
+
+// Overlaps samples in 'midBuffer' with the samples in 'pInputBuffer' at position
+// of 'ovlPos'.
+inline void TDStretch::overlap(SAMPLETYPE *pOutput, const SAMPLETYPE *pInput, uint ovlPos) const
+{
+    if (channels == 2) 
+    {
+        // stereo sound
+        overlapStereo(pOutput, pInput + 2 * ovlPos);
+    } else {
+        // mono sound.
+        overlapMono(pOutput, pInput + ovlPos);
+    }
+}
+
+
+
+// Seeks for the optimal overlap-mixing position. The 'stereo' version of the
+// routine
+//
+// The best position is determined as the position where the two overlapped
+// sample sequences are 'most alike', in terms of the highest cross-correlation
+// value over the overlapping period
+int TDStretch::seekBestOverlapPositionFull(const SAMPLETYPE *refPos) 
+{
+    int bestOffs;
+    double bestCorr, corr;
+    int i;
+
+    bestCorr = FLT_MIN;
+    bestOffs = 0;
+
+    // Scans for the best correlation value by testing each possible position
+    // over the permitted range.
+    for (i = 0; i < seekLength; i ++) 
+    {
+        // Calculates correlation value for the mixing position corresponding
+        // to 'i'
+        corr = calcCrossCorr(refPos + channels * i, pMidBuffer);
+        // heuristic rule to slightly favour values close to mid of the range
+        double tmp = (double)(2 * i - seekLength) / (double)seekLength;
+        corr = ((corr + 0.1) * (1.0 - 0.25 * tmp * tmp));
+
+        // Checks for the highest correlation value
+        if (corr > bestCorr) 
+        {
+            bestCorr = corr;
+            bestOffs = i;
+        }
+    }
+    // clear cross correlation routine state if necessary (is so e.g. in MMX routines).
+    clearCrossCorrState();
+
+    return bestOffs;
+}
+
+
+// Seeks for the optimal overlap-mixing position. The 'stereo' version of the
+// routine
+//
+// The best position is determined as the position where the two overlapped
+// sample sequences are 'most alike', in terms of the highest cross-correlation
+// value over the overlapping period
+int TDStretch::seekBestOverlapPositionQuick(const SAMPLETYPE *refPos) 
+{
+    int j;
+    int bestOffs;
+    double bestCorr, corr;
+    int scanCount, corrOffset, tempOffset;
+
+    bestCorr = FLT_MIN;
+    bestOffs = _scanOffsets[0][0];
+    corrOffset = 0;
+    tempOffset = 0;
+
+    // Scans for the best correlation value using four-pass hierarchical search.
+    //
+    // The look-up table 'scans' has hierarchical position adjusting steps.
+    // In first pass the routine searhes for the highest correlation with 
+    // relatively coarse steps, then rescans the neighbourhood of the highest
+    // correlation with better resolution and so on.
+    for (scanCount = 0;scanCount < 4; scanCount ++) 
+    {
+        j = 0;
+        while (_scanOffsets[scanCount][j]) 
+        {
+            tempOffset = corrOffset + _scanOffsets[scanCount][j];
+            if (tempOffset >= seekLength) break;
+
+            // Calculates correlation value for the mixing position corresponding
+            // to 'tempOffset'
+            corr = (double)calcCrossCorr(refPos + channels * tempOffset, pMidBuffer);
+            // heuristic rule to slightly favour values close to mid of the range
+            double tmp = (double)(2 * tempOffset - seekLength) / seekLength;
+            corr = ((corr + 0.1) * (1.0 - 0.25 * tmp * tmp));
+
+            // Checks for the highest correlation value
+            if (corr > bestCorr) 
+            {
+                bestCorr = corr;
+                bestOffs = tempOffset;
+            }
+            j ++;
+        }
+        corrOffset = bestOffs;
+    }
+    // clear cross correlation routine state if necessary (is so e.g. in MMX routines).
+    clearCrossCorrState();
+
+    return bestOffs;
+}
+
+
+
+/// clear cross correlation routine state if necessary 
+void TDStretch::clearCrossCorrState()
+{
+    // default implementation is empty.
+}
+
+
+/// Calculates processing sequence length according to tempo setting
+void TDStretch::calcSeqParameters()
+{
+    // Adjust tempo param according to tempo, so that variating processing sequence length is used
+    // at varius tempo settings, between the given low...top limits
+    #define AUTOSEQ_TEMPO_LOW   0.5     // auto setting low tempo range (-50%)
+    #define AUTOSEQ_TEMPO_TOP   2.0     // auto setting top tempo range (+100%)
+
+    // sequence-ms setting values at above low & top tempo
+    #define AUTOSEQ_AT_MIN      125.0
+    #define AUTOSEQ_AT_MAX      50.0
+    #define AUTOSEQ_K           ((AUTOSEQ_AT_MAX - AUTOSEQ_AT_MIN) / (AUTOSEQ_TEMPO_TOP - AUTOSEQ_TEMPO_LOW))
+    #define AUTOSEQ_C           (AUTOSEQ_AT_MIN - (AUTOSEQ_K) * (AUTOSEQ_TEMPO_LOW))
+
+    // seek-window-ms setting values at above low & top tempo
+    #define AUTOSEEK_AT_MIN     25.0
+    #define AUTOSEEK_AT_MAX     15.0
+    #define AUTOSEEK_K          ((AUTOSEEK_AT_MAX - AUTOSEEK_AT_MIN) / (AUTOSEQ_TEMPO_TOP - AUTOSEQ_TEMPO_LOW))
+    #define AUTOSEEK_C          (AUTOSEEK_AT_MIN - (AUTOSEEK_K) * (AUTOSEQ_TEMPO_LOW))
+
+    #define CHECK_LIMITS(x, mi, ma) (((x) < (mi)) ? (mi) : (((x) > (ma)) ? (ma) : (x)))
+
+    double seq, seek;
+    
+    if (bAutoSeqSetting)
+    {
+        seq = AUTOSEQ_C + AUTOSEQ_K * tempo;
+        seq = CHECK_LIMITS(seq, AUTOSEQ_AT_MAX, AUTOSEQ_AT_MIN);
+        sequenceMs = (int)(seq + 0.5);
+    }
+
+    if (bAutoSeekSetting)
+    {
+        seek = AUTOSEEK_C + AUTOSEEK_K * tempo;
+        seek = CHECK_LIMITS(seek, AUTOSEEK_AT_MAX, AUTOSEEK_AT_MIN);
+        seekWindowMs = (int)(seek + 0.5);
+    }
+
+    // Update seek window lengths
+    seekWindowLength = (sampleRate * sequenceMs) / 1000;
+    if (seekWindowLength < 2 * overlapLength) 
+    {
+        seekWindowLength = 2 * overlapLength;
+    }
+    seekLength = (sampleRate * seekWindowMs) / 1000;
+}
+
+
+
+// Sets new target tempo. Normal tempo = 'SCALE', smaller values represent slower 
+// tempo, larger faster tempo.
+void TDStretch::setTempo(float newTempo)
+{
+    int intskip;
+
+    tempo = newTempo;
+
+    // Calculate new sequence duration
+    calcSeqParameters();
+
+    // Calculate ideal skip length (according to tempo value) 
+    nominalSkip = tempo * (seekWindowLength - overlapLength);
+    intskip = (int)(nominalSkip + 0.5f);
+
+    // Calculate how many samples are needed in the 'inputBuffer' to 
+    // process another batch of samples
+    //sampleReq = max(intskip + overlapLength, seekWindowLength) + seekLength / 2;
+    sampleReq = max(intskip + overlapLength, seekWindowLength) + seekLength;
+}
+
+
+
+// Sets the number of channels, 1 = mono, 2 = stereo
+void TDStretch::setChannels(int numChannels)
+{
+    assert(numChannels > 0);
+    if (channels == numChannels) return;
+    assert(numChannels == 1 || numChannels == 2);
+
+    channels = numChannels;
+    inputBuffer.setChannels(channels);
+    outputBuffer.setChannels(channels);
+}
+
+
+// nominal tempo, no need for processing, just pass the samples through
+// to outputBuffer
+/*
+void TDStretch::processNominalTempo()
+{
+    assert(tempo == 1.0f);
+
+    if (bMidBufferDirty) 
+    {
+        // If there are samples in pMidBuffer waiting for overlapping,
+        // do a single sliding overlapping with them in order to prevent a 
+        // clicking distortion in the output sound
+        if (inputBuffer.numSamples() < overlapLength) 
+        {
+            // wait until we've got overlapLength input samples
+            return;
+        }
+        // Mix the samples in the beginning of 'inputBuffer' with the 
+        // samples in 'midBuffer' using sliding overlapping 
+        overlap(outputBuffer.ptrEnd(overlapLength), inputBuffer.ptrBegin(), 0);
+        outputBuffer.putSamples(overlapLength);
+        inputBuffer.receiveSamples(overlapLength);
+        clearMidBuffer();
+        // now we've caught the nominal sample flow and may switch to
+        // bypass mode
+    }
+
+    // Simply bypass samples from input to output
+    outputBuffer.moveSamples(inputBuffer);
+}
+*/
+
+#include <stdio.h>
+
+// Processes as many processing frames of the samples 'inputBuffer', store
+// the result into 'outputBuffer'
+void TDStretch::processSamples()
+{
+    int ovlSkip, offset;
+    int temp;
+
+    /* Removed this small optimization - can introduce a click to sound when tempo setting
+       crosses the nominal value
+    if (tempo == 1.0f) 
+    {
+        // tempo not changed from the original, so bypass the processing
+        processNominalTempo();
+        return;
+    }
+    */
+
+    // Process samples as long as there are enough samples in 'inputBuffer'
+    // to form a processing frame.
+    while ((int)inputBuffer.numSamples() >= sampleReq) 
+    {
+        // If tempo differs from the normal ('SCALE'), scan for the best overlapping
+        // position
+        offset = seekBestOverlapPosition(inputBuffer.ptrBegin());
+
+        // Mix the samples in the 'inputBuffer' at position of 'offset' with the 
+        // samples in 'midBuffer' using sliding overlapping
+        // ... first partially overlap with the end of the previous sequence
+        // (that's in 'midBuffer')
+        overlap(outputBuffer.ptrEnd((uint)overlapLength), inputBuffer.ptrBegin(), (uint)offset);
+        outputBuffer.putSamples((uint)overlapLength);
+
+        // ... then copy sequence samples from 'inputBuffer' to output:
+
+        // length of sequence
+        temp = (seekWindowLength - 2 * overlapLength);
+
+        // crosscheck that we don't have buffer overflow...
+        if ((int)inputBuffer.numSamples() < (offset + temp + overlapLength * 2))
+        {
+            continue;    // just in case, shouldn't really happen
+        }
+
+        outputBuffer.putSamples(inputBuffer.ptrBegin() + channels * (offset + overlapLength), (uint)temp);
+
+        // Copies the end of the current sequence from 'inputBuffer' to 
+        // 'midBuffer' for being mixed with the beginning of the next 
+        // processing sequence and so on
+        assert((offset + temp + overlapLength * 2) <= (int)inputBuffer.numSamples());
+        memcpy(pMidBuffer, inputBuffer.ptrBegin() + channels * (offset + temp + overlapLength), 
+            channels * sizeof(SAMPLETYPE) * overlapLength);
+
+        // Remove the processed samples from the input buffer. Update
+        // the difference between integer & nominal skip step to 'skipFract'
+        // in order to prevent the error from accumulating over time.
+        skipFract += nominalSkip;   // real skip size
+        ovlSkip = (int)skipFract;   // rounded to integer skip
+        skipFract -= ovlSkip;       // maintain the fraction part, i.e. real vs. integer skip
+        inputBuffer.receiveSamples((uint)ovlSkip);
+    }
+}
+
+
+// Adds 'numsamples' pcs of samples from the 'samples' memory position into
+// the input of the object.
+void TDStretch::putSamples(const SAMPLETYPE *samples, uint nSamples)
+{
+    // Add the samples into the input buffer
+    inputBuffer.putSamples(samples, nSamples);
+    // Process the samples in input buffer
+    processSamples();
+}
+
+
+
+/// Set new overlap length parameter & reallocate RefMidBuffer if necessary.
+void TDStretch::acceptNewOverlapLength(int newOverlapLength)
+{
+    int prevOvl;
+
+    assert(newOverlapLength >= 0);
+    prevOvl = overlapLength;
+    overlapLength = newOverlapLength;
+
+    if (overlapLength > prevOvl)
+    {
+        delete[] pMidBufferUnaligned;
+
+        pMidBufferUnaligned = new SAMPLETYPE[overlapLength * 2 + 16 / sizeof(SAMPLETYPE)];
+        // ensure that 'pMidBuffer' is aligned to 16 byte boundary for efficiency
+        pMidBuffer = (SAMPLETYPE *)SOUNDTOUCH_ALIGN_POINTER_16(pMidBufferUnaligned);
+
+        clearMidBuffer();
+    }
+}
+
+
+// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+// depending on if we've a MMX/SSE/etc-capable CPU available or not.
+void * TDStretch::operator new(size_t s)
+{
+    // Notice! don't use "new TDStretch" directly, use "newInstance" to create a new instance instead!
+    ST_THROW_RT_ERROR("Error in TDStretch::new: Don't use 'new TDStretch' directly, use 'newInstance' member instead!");
+    return newInstance();
+}
+
+
+TDStretch * TDStretch::newInstance()
+{
+    uint uExtensions;
+
+    uExtensions = detectCPUextensions();
+
+    // Check if MMX/SSE instruction set extensions supported by CPU
+
+#ifdef SOUNDTOUCH_ALLOW_MMX
+    // MMX routines available only with integer sample types
+    if (uExtensions & SUPPORT_MMX)
+    {
+        return ::new TDStretchMMX;
+    }
+    else
+#endif // SOUNDTOUCH_ALLOW_MMX
+
+
+#ifdef SOUNDTOUCH_ALLOW_SSE
+    if (uExtensions & SUPPORT_SSE)
+    {
+        // SSE support
+        return ::new TDStretchSSE;
+    }
+    else
+#endif // SOUNDTOUCH_ALLOW_SSE
+
+    {
+        // ISA optimizations not supported, use plain C version
+        return ::new TDStretch;
+    }
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Integer arithmetics specific algorithm implementations.
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#ifdef SOUNDTOUCH_INTEGER_SAMPLES
+
+// Overlaps samples in 'midBuffer' with the samples in 'input'. The 'Stereo' 
+// version of the routine.
+void TDStretch::overlapStereo(short *poutput, const short *input) const
+{
+    int i;
+    short temp;
+    int cnt2;
+
+    for (i = 0; i < overlapLength ; i ++) 
+    {
+        temp = (short)(overlapLength - i);
+        cnt2 = 2 * i;
+        poutput[cnt2] = (input[cnt2] * i + pMidBuffer[cnt2] * temp )  / overlapLength;
+        poutput[cnt2 + 1] = (input[cnt2 + 1] * i + pMidBuffer[cnt2 + 1] * temp ) / overlapLength;
+    }
+}
+
+// Calculates the x having the closest 2^x value for the given value
+static int _getClosest2Power(double value)
+{
+    return (int)(log(value) / log(2.0) + 0.5);
+}
+
+
+/// Calculates overlap period length in samples.
+/// Integer version rounds overlap length to closest power of 2
+/// for a divide scaling operation.
+void TDStretch::calculateOverlapLength(int aoverlapMs)
+{
+    int newOvl;
+
+    assert(aoverlapMs >= 0);
+
+    // calculate overlap length so that it's power of 2 - thus it's easy to do
+    // integer division by right-shifting. Term "-1" at end is to account for 
+    // the extra most significatnt bit left unused in result by signed multiplication 
+    overlapDividerBits = _getClosest2Power((sampleRate * aoverlapMs) / 1000.0) - 1;
+    if (overlapDividerBits > 9) overlapDividerBits = 9;
+    if (overlapDividerBits < 3) overlapDividerBits = 3;
+    newOvl = (int)pow(2.0, (int)overlapDividerBits + 1);    // +1 => account for -1 above
+
+    acceptNewOverlapLength(newOvl);
+
+    // calculate sloping divider so that crosscorrelation operation won't 
+    // overflow 32-bit register. Max. sum of the crosscorrelation sum without 
+    // divider would be 2^30*(N^3-N)/3, where N = overlap length
+    slopingDivider = (newOvl * newOvl - 1) / 3;
+}
+
+
+double TDStretch::calcCrossCorr(const short *mixingPos, const short *compare) const
+{
+    long corr;
+    long norm;
+    int i;
+
+    corr = norm = 0;
+    // Same routine for stereo and mono. For stereo, unroll loop for better
+    // efficiency and gives slightly better resolution against rounding. 
+    // For mono it same routine, just  unrolls loop by factor of 4
+    for (i = 0; i < channels * overlapLength; i += 4) 
+    {
+        corr += (mixingPos[i] * compare[i] + 
+                 mixingPos[i + 1] * compare[i + 1] +
+                 mixingPos[i + 2] * compare[i + 2] + 
+                 mixingPos[i + 3] * compare[i + 3]) >> overlapDividerBits;
+        norm += (mixingPos[i] * mixingPos[i] + 
+                 mixingPos[i + 1] * mixingPos[i + 1] +
+                 mixingPos[i + 2] * mixingPos[i + 2] + 
+                 mixingPos[i + 3] * mixingPos[i + 3]) >> overlapDividerBits;
+    }
+
+    // Normalize result by dividing by sqrt(norm) - this step is easiest 
+    // done using floating point operation
+    if (norm == 0) norm = 1;    // to avoid div by zero
+    return (double)corr / sqrt((double)norm);
+}
+
+#endif // SOUNDTOUCH_INTEGER_SAMPLES
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Floating point arithmetics specific algorithm implementations.
+//
+
+#ifdef SOUNDTOUCH_FLOAT_SAMPLES
+
+// Overlaps samples in 'midBuffer' with the samples in 'pInput'
+void TDStretch::overlapStereo(float *pOutput, const float *pInput) const
+{
+    int i;
+    float fScale;
+    float f1;
+    float f2;
+
+    fScale = 1.0f / (float)overlapLength;
+
+    f1 = 0;
+    f2 = 1.0f;
+
+    for (i = 0; i < 2 * (int)overlapLength ; i += 2) 
+    {
+        pOutput[i + 0] = pInput[i + 0] * f1 + pMidBuffer[i + 0] * f2;
+        pOutput[i + 1] = pInput[i + 1] * f1 + pMidBuffer[i + 1] * f2;
+
+        f1 += fScale;
+        f2 -= fScale;
+    }
+}
+
+
+/// Calculates overlapInMsec period length in samples.
+void TDStretch::calculateOverlapLength(int overlapInMsec)
+{
+    int newOvl;
+
+    assert(overlapInMsec >= 0);
+    newOvl = (sampleRate * overlapInMsec) / 1000;
+    if (newOvl < 16) newOvl = 16;
+
+    // must be divisible by 8
+    newOvl -= newOvl % 8;
+
+    acceptNewOverlapLength(newOvl);
+}
+
+
+double TDStretch::calcCrossCorr(const float *mixingPos, const float *compare) const
+{
+    double corr;
+    double norm;
+    int i;
+
+    corr = norm = 0;
+    // Same routine for stereo and mono. For Stereo, unroll by factor of 2.
+    // For mono it's same routine yet unrollsd by factor of 4.
+    for (i = 0; i < channels * overlapLength; i += 4) 
+    {
+        corr += mixingPos[i] * compare[i] +
+                mixingPos[i + 1] * compare[i + 1];
+
+        norm += mixingPos[i] * mixingPos[i] + 
+                mixingPos[i + 1] * mixingPos[i + 1];
+
+        // unroll the loop for better CPU efficiency:
+        corr += mixingPos[i + 2] * compare[i + 2] +
+                mixingPos[i + 3] * compare[i + 3];
+
+        norm += mixingPos[i + 2] * mixingPos[i + 2] +
+                mixingPos[i + 3] * mixingPos[i + 3];
+    }
+
+    if (norm < 1e-9) norm = 1.0;    // to avoid div by zero
+    return corr / sqrt(norm);
+}
+
+#endif // SOUNDTOUCH_FLOAT_SAMPLES

Externals/soundtouch/TDStretch.h

@@ -0,0 +1,268 @@
+////////////////////////////////////////////////////////////////////////////////
+/// 
+/// Sampled sound tempo changer/time stretch algorithm. Changes the sound tempo 
+/// while maintaining the original pitch by using a time domain WSOLA-like method 
+/// with several performance-increasing tweaks.
+///
+/// Note : MMX/SSE optimized functions reside in separate, platform-specific files 
+/// 'mmx_optimized.cpp' and 'sse_optimized.cpp'
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-04-01 22:49:30 +0300 (Sun, 01 Apr 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: TDStretch.h 137 2012-04-01 19:49:30Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef TDStretch_H
+#define TDStretch_H
+
+#include <stddef.h>
+#include "STTypes.h"
+#include "RateTransposer.h"
+#include "FIFOSamplePipe.h"
+
+namespace soundtouch
+{
+
+/// Default values for sound processing parameters:
+/// Notice that the default parameters are tuned for contemporary popular music 
+/// processing. For speech processing applications these parameters suit better:
+///     #define DEFAULT_SEQUENCE_MS     40
+///     #define DEFAULT_SEEKWINDOW_MS   15
+///     #define DEFAULT_OVERLAP_MS      8
+///
+
+/// Default length of a single processing sequence, in milliseconds. This determines to how 
+/// long sequences the original sound is chopped in the time-stretch algorithm.
+///
+/// The larger this value is, the lesser sequences are used in processing. In principle
+/// a bigger value sounds better when slowing down tempo, but worse when increasing tempo
+/// and vice versa.
+///
+/// Increasing this value reduces computational burden & vice versa.
+//#define DEFAULT_SEQUENCE_MS         40
+#define DEFAULT_SEQUENCE_MS         USE_AUTO_SEQUENCE_LEN
+
+/// Giving this value for the sequence length sets automatic parameter value
+/// according to tempo setting (recommended)
+#define USE_AUTO_SEQUENCE_LEN       0
+
+/// Seeking window default length in milliseconds for algorithm that finds the best possible 
+/// overlapping location. This determines from how wide window the algorithm may look for an 
+/// optimal joining location when mixing the sound sequences back together. 
+///
+/// The bigger this window setting is, the higher the possibility to find a better mixing
+/// position will become, but at the same time large values may cause a "drifting" artifact
+/// because consequent sequences will be taken at more uneven intervals.
+///
+/// If there's a disturbing artifact that sounds as if a constant frequency was drifting 
+/// around, try reducing this setting.
+///
+/// Increasing this value increases computational burden & vice versa.
+//#define DEFAULT_SEEKWINDOW_MS       15
+#define DEFAULT_SEEKWINDOW_MS       USE_AUTO_SEEKWINDOW_LEN
+
+/// Giving this value for the seek window length sets automatic parameter value
+/// according to tempo setting (recommended)
+#define USE_AUTO_SEEKWINDOW_LEN     0
+
+/// Overlap length in milliseconds. When the chopped sound sequences are mixed back together, 
+/// to form a continuous sound stream, this parameter defines over how long period the two 
+/// consecutive sequences are let to overlap each other. 
+///
+/// This shouldn't be that critical parameter. If you reduce the DEFAULT_SEQUENCE_MS setting 
+/// by a large amount, you might wish to try a smaller value on this.
+///
+/// Increasing this value increases computational burden & vice versa.
+#define DEFAULT_OVERLAP_MS      8
+
+
+/// Class that does the time-stretch (tempo change) effect for the processed
+/// sound.
+class TDStretch : public FIFOProcessor
+{
+protected:
+    int channels;
+    int sampleReq;
+    float tempo;
+
+    SAMPLETYPE *pMidBuffer;
+    SAMPLETYPE *pMidBufferUnaligned;
+    int overlapLength;
+    int seekLength;
+    int seekWindowLength;
+    int overlapDividerBits;
+    int slopingDivider;
+    float nominalSkip;
+    float skipFract;
+    FIFOSampleBuffer outputBuffer;
+    FIFOSampleBuffer inputBuffer;
+    BOOL bQuickSeek;
+
+    int sampleRate;
+    int sequenceMs;
+    int seekWindowMs;
+    int overlapMs;
+    BOOL bAutoSeqSetting;
+    BOOL bAutoSeekSetting;
+
+    void acceptNewOverlapLength(int newOverlapLength);
+
+    virtual void clearCrossCorrState();
+    void calculateOverlapLength(int overlapMs);
+
+    virtual double calcCrossCorr(const SAMPLETYPE *mixingPos, const SAMPLETYPE *compare) const;
+
+    virtual int seekBestOverlapPositionFull(const SAMPLETYPE *refPos);
+    virtual int seekBestOverlapPositionQuick(const SAMPLETYPE *refPos);
+    int seekBestOverlapPosition(const SAMPLETYPE *refPos);
+
+    virtual void overlapStereo(SAMPLETYPE *output, const SAMPLETYPE *input) const;
+    virtual void overlapMono(SAMPLETYPE *output, const SAMPLETYPE *input) const;
+
+    void clearMidBuffer();
+    void overlap(SAMPLETYPE *output, const SAMPLETYPE *input, uint ovlPos) const;
+
+    void calcSeqParameters();
+
+    /// Changes the tempo of the given sound samples.
+    /// Returns amount of samples returned in the "output" buffer.
+    /// The maximum amount of samples that can be returned at a time is set by
+    /// the 'set_returnBuffer_size' function.
+    void processSamples();
+    
+public:
+    TDStretch();
+    virtual ~TDStretch();
+
+    /// Operator 'new' is overloaded so that it automatically creates a suitable instance 
+    /// depending on if we've a MMX/SSE/etc-capable CPU available or not.
+    static void *operator new(size_t s);
+
+    /// Use this function instead of "new" operator to create a new instance of this class. 
+    /// This function automatically chooses a correct feature set depending on if the CPU
+    /// supports MMX/SSE/etc extensions.
+    static TDStretch *newInstance();
+    
+    /// Returns the output buffer object
+    FIFOSamplePipe *getOutput() { return &outputBuffer; };
+
+    /// Returns the input buffer object
+    FIFOSamplePipe *getInput() { return &inputBuffer; };
+
+    /// Sets new target tempo. Normal tempo = 'SCALE', smaller values represent slower 
+    /// tempo, larger faster tempo.
+    void setTempo(float newTempo);
+
+    /// Returns nonzero if there aren't any samples available for outputting.
+    virtual void clear();
+
+    /// Clears the input buffer
+    void clearInput();
+
+    /// Sets the number of channels, 1 = mono, 2 = stereo
+    void setChannels(int numChannels);
+
+    /// Enables/disables the quick position seeking algorithm. Zero to disable, 
+    /// nonzero to enable
+    void enableQuickSeek(BOOL enable);
+
+    /// Returns nonzero if the quick seeking algorithm is enabled.
+    BOOL isQuickSeekEnabled() const;
+
+    /// Sets routine control parameters. These control are certain time constants
+    /// defining how the sound is stretched to the desired duration.
+    //
+    /// 'sampleRate' = sample rate of the sound
+    /// 'sequenceMS' = one processing sequence length in milliseconds
+    /// 'seekwindowMS' = seeking window length for scanning the best overlapping 
+    ///      position
+    /// 'overlapMS' = overlapping length
+    void setParameters(int sampleRate,          ///< Samplerate of sound being processed (Hz)
+                       int sequenceMS = -1,     ///< Single processing sequence length (ms)
+                       int seekwindowMS = -1,   ///< Offset seeking window length (ms)
+                       int overlapMS = -1       ///< Sequence overlapping length (ms)
+                       );
+
+    /// Get routine control parameters, see setParameters() function.
+    /// Any of the parameters to this function can be NULL, in such case corresponding parameter
+    /// value isn't returned.
+    void getParameters(int *pSampleRate, int *pSequenceMs, int *pSeekWindowMs, int *pOverlapMs) const;
+
+    /// Adds 'numsamples' pcs of samples from the 'samples' memory position into
+    /// the input of the object.
+    virtual void putSamples(
+            const SAMPLETYPE *samples,  ///< Input sample data
+            uint numSamples                         ///< Number of samples in 'samples' so that one sample
+                                                    ///< contains both channels if stereo
+            );
+
+    /// return nominal input sample requirement for triggering a processing batch
+    int getInputSampleReq() const
+    {
+        return (int)(nominalSkip + 0.5);
+    }
+
+    /// return nominal output sample amount when running a processing batch
+    int getOutputBatchSize() const
+    {
+        return seekWindowLength - overlapLength;
+    }
+};
+
+
+
+// Implementation-specific class declarations:
+
+#ifdef SOUNDTOUCH_ALLOW_MMX
+    /// Class that implements MMX optimized routines for 16bit integer samples type.
+    class TDStretchMMX : public TDStretch
+    {
+    protected:
+        double calcCrossCorr(const short *mixingPos, const short *compare) const;
+        virtual void overlapStereo(short *output, const short *input) const;
+        virtual void clearCrossCorrState();
+    };
+#endif /// SOUNDTOUCH_ALLOW_MMX
+
+
+#ifdef SOUNDTOUCH_ALLOW_SSE
+    /// Class that implements SSE optimized routines for floating point samples type.
+    class TDStretchSSE : public TDStretch
+    {
+    protected:
+        double calcCrossCorr(const float *mixingPos, const float *compare) const;
+    };
+
+#endif /// SOUNDTOUCH_ALLOW_SSE
+
+}
+#endif  /// TDStretch_H

Externals/soundtouch/cpu_detect.h

@@ -0,0 +1,62 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// A header file for detecting the Intel MMX instructions set extension.
+///
+/// Please see 'mmx_win.cpp', 'mmx_cpp.cpp' and 'mmx_non_x86.cpp' for the 
+/// routine implementations for x86 Windows, x86 gnu version and non-x86 
+/// platforms, respectively.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2008-02-10 18:26:55 +0200 (Sun, 10 Feb 2008) $
+// File revision : $Revision: 4 $
+//
+// $Id: cpu_detect.h 11 2008-02-10 16:26:55Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#ifndef _CPU_DETECT_H_
+#define _CPU_DETECT_H_
+
+#include "STTypes.h"
+
+#define SUPPORT_MMX         0x0001
+#define SUPPORT_3DNOW       0x0002
+#define SUPPORT_ALTIVEC     0x0004
+#define SUPPORT_SSE         0x0008
+#define SUPPORT_SSE2        0x0010
+
+/// Checks which instruction set extensions are supported by the CPU.
+///
+/// \return A bitmask of supported extensions, see SUPPORT_... defines.
+uint detectCPUextensions(void);
+
+/// Disables given set of instruction extensions. See SUPPORT_... defines.
+void disableExtensions(uint wDisableMask);
+
+#endif  // _CPU_DETECT_H_

Externals/soundtouch/cpu_detect_x86.cpp

@@ -0,0 +1,137 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// Generic version of the x86 CPU extension detection routine.
+///
+/// This file is for GNU & other non-Windows compilers, see 'cpu_detect_x86_win.cpp' 
+/// for the Microsoft compiler version.
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-11-08 20:44:37 +0200 (Thu, 08 Nov 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: cpu_detect_x86.cpp 159 2012-11-08 18:44:37Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include "cpu_detect.h"
+#include "STTypes.h"
+
+#if defined(SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS)
+
+    #if defined(__GNUC__) && defined(__i386__)
+        // gcc
+        #include "cpuid.h"
+    #elif defined(_M_IX86)
+        // windows non-gcc
+        #include <intrin.h>
+    #endif
+
+	#define bit_MMX     (1 << 23)
+	#define bit_SSE     (1 << 25)
+	#define bit_SSE2    (1 << 26)
+#endif
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// processor instructions extension detection routines
+//
+//////////////////////////////////////////////////////////////////////////////
+
+// Flag variable indicating whick ISA extensions are disabled (for debugging)
+static uint _dwDisabledISA = 0x00;      // 0xffffffff; //<- use this to disable all extensions
+
+// Disables given set of instruction extensions. See SUPPORT_... defines.
+void disableExtensions(uint dwDisableMask)
+{
+    _dwDisabledISA = dwDisableMask;
+}
+
+
+
+/// Checks which instruction set extensions are supported by the CPU.
+uint detectCPUextensions(void)
+{
+/// If building for a 64bit system (no Itanium) and the user wants optimizations.
+/// Return the OR of SUPPORT_{MMX,SSE,SSE2}. 11001 or 0x19.
+/// Keep the _dwDisabledISA test (2 more operations, could be eliminated).
+#if ((defined(__GNUC__) && defined(__x86_64__)) \
+    || defined(_M_X64))  \
+    && defined(SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS)
+    return 0x19 & ~_dwDisabledISA;
+
+/// If building for a 32bit system and the user wants optimizations.
+/// Keep the _dwDisabledISA test (2 more operations, could be eliminated).
+#elif ((defined(__GNUC__) && defined(__i386__)) \
+    || defined(_M_IX86))  \
+    && defined(SOUNDTOUCH_ALLOW_X86_OPTIMIZATIONS)
+
+    if (_dwDisabledISA == 0xffffffff) return 0;
+ 
+    uint res = 0;
+ 
+#if defined(__GNUC__)
+    // GCC version of cpuid. Requires GCC 4.3.0 or later for __cpuid intrinsic support.
+    uint eax, ebx, ecx, edx;  // unsigned int is the standard type. uint is defined by the compiler and not guaranteed to be portable.
+
+    // Check if no cpuid support.
+    if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx)) return 0; // always disable extensions.
+
+    if (edx & bit_MMX)  res = res | SUPPORT_MMX;
+    if (edx & bit_SSE)  res = res | SUPPORT_SSE;
+    if (edx & bit_SSE2) res = res | SUPPORT_SSE2;
+
+#else
+    // Window / VS version of cpuid. Notice that Visual Studio 2005 or later required 
+    // for __cpuid intrinsic support.
+    int reg[4] = {-1};
+
+    // Check if no cpuid support.
+    __cpuid(reg,0);
+    if ((unsigned int)reg[0] == 0) return 0; // always disable extensions.
+
+    __cpuid(reg,1);
+    if ((unsigned int)reg[3] & bit_MMX)  res = res | SUPPORT_MMX;
+    if ((unsigned int)reg[3] & bit_SSE)  res = res | SUPPORT_SSE;
+    if ((unsigned int)reg[3] & bit_SSE2) res = res | SUPPORT_SSE2;
+
+#endif
+
+    return res & ~_dwDisabledISA;
+
+#else
+
+/// One of these is true:
+/// 1) We don't want optimizations.
+/// 2) Using an unsupported compiler.
+/// 3) Running on a non-x86 platform.
+    return 0;
+
+#endif
+}

Externals/soundtouch/mmx_optimized.cpp

@@ -0,0 +1,317 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// MMX optimized routines. All MMX optimized functions have been gathered into 
+/// this single source code file, regardless to their class or original source 
+/// code file, in order to ease porting the library to other compiler and 
+/// processor platforms.
+///
+/// The MMX-optimizations are programmed using MMX compiler intrinsics that
+/// are supported both by Microsoft Visual C++ and GCC compilers, so this file
+/// should compile with both toolsets.
+///
+/// NOTICE: If using Visual Studio 6.0, you'll need to install the "Visual C++ 
+/// 6.0 processor pack" update to support compiler intrinsic syntax. The update
+/// is available for download at Microsoft Developers Network, see here:
+/// http://msdn.microsoft.com/en-us/vstudio/aa718349.aspx
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-11-08 20:53:01 +0200 (Thu, 08 Nov 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: mmx_optimized.cpp 160 2012-11-08 18:53:01Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include "STTypes.h"
+
+#ifdef SOUNDTOUCH_ALLOW_MMX
+// MMX routines available only with integer sample type
+
+using namespace soundtouch;
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of MMX optimized functions of class 'TDStretchMMX'
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "TDStretch.h"
+#include <mmintrin.h>
+#include <limits.h>
+#include <math.h>
+
+
+// Calculates cross correlation of two buffers
+double TDStretchMMX::calcCrossCorr(const short *pV1, const short *pV2) const
+{
+    const __m64 *pVec1, *pVec2;
+    __m64 shifter;
+    __m64 accu, normaccu;
+    long corr, norm;
+    int i;
+   
+    pVec1 = (__m64*)pV1;
+    pVec2 = (__m64*)pV2;
+
+    shifter = _m_from_int(overlapDividerBits);
+    normaccu = accu = _mm_setzero_si64();
+
+    // Process 4 parallel sets of 2 * stereo samples or 4 * mono samples 
+    // during each round for improved CPU-level parallellization.
+    for (i = 0; i < channels * overlapLength / 16; i ++)
+    {
+        __m64 temp, temp2;
+
+        // dictionary of instructions:
+        // _m_pmaddwd   : 4*16bit multiply-add, resulting two 32bits = [a0*b0+a1*b1 ; a2*b2+a3*b3]
+        // _mm_add_pi32 : 2*32bit add
+        // _m_psrad     : 32bit right-shift
+
+        temp = _mm_add_pi32(_mm_madd_pi16(pVec1[0], pVec2[0]),
+                            _mm_madd_pi16(pVec1[1], pVec2[1]));
+        temp2 = _mm_add_pi32(_mm_madd_pi16(pVec1[0], pVec1[0]),
+                             _mm_madd_pi16(pVec1[1], pVec1[1]));
+        accu = _mm_add_pi32(accu, _mm_sra_pi32(temp, shifter));
+        normaccu = _mm_add_pi32(normaccu, _mm_sra_pi32(temp2, shifter));
+
+        temp = _mm_add_pi32(_mm_madd_pi16(pVec1[2], pVec2[2]),
+                            _mm_madd_pi16(pVec1[3], pVec2[3]));
+        temp2 = _mm_add_pi32(_mm_madd_pi16(pVec1[2], pVec1[2]),
+                             _mm_madd_pi16(pVec1[3], pVec1[3]));
+        accu = _mm_add_pi32(accu, _mm_sra_pi32(temp, shifter));
+        normaccu = _mm_add_pi32(normaccu, _mm_sra_pi32(temp2, shifter));
+
+        pVec1 += 4;
+        pVec2 += 4;
+    }
+
+    // copy hi-dword of mm0 to lo-dword of mm1, then sum mmo+mm1
+    // and finally store the result into the variable "corr"
+
+    accu = _mm_add_pi32(accu, _mm_srli_si64(accu, 32));
+    corr = _m_to_int(accu);
+
+    normaccu = _mm_add_pi32(normaccu, _mm_srli_si64(normaccu, 32));
+    norm = _m_to_int(normaccu);
+
+    // Clear MMS state
+    _m_empty();
+
+    // Normalize result by dividing by sqrt(norm) - this step is easiest 
+    // done using floating point operation
+    if (norm == 0) norm = 1;    // to avoid div by zero
+
+    return (double)corr / sqrt((double)norm);
+    // Note: Warning about the missing EMMS instruction is harmless
+    // as it'll be called elsewhere.
+}
+
+
+
+void TDStretchMMX::clearCrossCorrState()
+{
+    // Clear MMS state
+    _m_empty();
+    //_asm EMMS;
+}
+
+
+
+// MMX-optimized version of the function overlapStereo
+void TDStretchMMX::overlapStereo(short *output, const short *input) const
+{
+    const __m64 *pVinput, *pVMidBuf;
+    __m64 *pVdest;
+    __m64 mix1, mix2, adder, shifter;
+    int i;
+
+    pVinput  = (const __m64*)input;
+    pVMidBuf = (const __m64*)pMidBuffer;
+    pVdest   = (__m64*)output;
+
+    // mix1  = mixer values for 1st stereo sample
+    // mix1  = mixer values for 2nd stereo sample
+    // adder = adder for updating mixer values after each round
+    
+    mix1  = _mm_set_pi16(0, overlapLength,   0, overlapLength);
+    adder = _mm_set_pi16(1, -1, 1, -1);
+    mix2  = _mm_add_pi16(mix1, adder);
+    adder = _mm_add_pi16(adder, adder);
+
+    // Overlaplength-division by shifter. "+1" is to account for "-1" deduced in
+    // overlapDividerBits calculation earlier.
+    shifter = _m_from_int(overlapDividerBits + 1);
+
+    for (i = 0; i < overlapLength / 4; i ++)
+    {
+        __m64 temp1, temp2;
+                
+        // load & shuffle data so that input & mixbuffer data samples are paired
+        temp1 = _mm_unpacklo_pi16(pVMidBuf[0], pVinput[0]);     // = i0l m0l i0r m0r
+        temp2 = _mm_unpackhi_pi16(pVMidBuf[0], pVinput[0]);     // = i1l m1l i1r m1r
+
+        // temp = (temp .* mix) >> shifter
+        temp1 = _mm_sra_pi32(_mm_madd_pi16(temp1, mix1), shifter);
+        temp2 = _mm_sra_pi32(_mm_madd_pi16(temp2, mix2), shifter);
+        pVdest[0] = _mm_packs_pi32(temp1, temp2); // pack 2*2*32bit => 4*16bit
+
+        // update mix += adder
+        mix1 = _mm_add_pi16(mix1, adder);
+        mix2 = _mm_add_pi16(mix2, adder);
+
+        // --- second round begins here ---
+
+        // load & shuffle data so that input & mixbuffer data samples are paired
+        temp1 = _mm_unpacklo_pi16(pVMidBuf[1], pVinput[1]);       // = i2l m2l i2r m2r
+        temp2 = _mm_unpackhi_pi16(pVMidBuf[1], pVinput[1]);       // = i3l m3l i3r m3r
+
+        // temp = (temp .* mix) >> shifter
+        temp1 = _mm_sra_pi32(_mm_madd_pi16(temp1, mix1), shifter);
+        temp2 = _mm_sra_pi32(_mm_madd_pi16(temp2, mix2), shifter);
+        pVdest[1] = _mm_packs_pi32(temp1, temp2); // pack 2*2*32bit => 4*16bit
+
+        // update mix += adder
+        mix1 = _mm_add_pi16(mix1, adder);
+        mix2 = _mm_add_pi16(mix2, adder);
+
+        pVinput  += 2;
+        pVMidBuf += 2;
+        pVdest   += 2;
+    }
+
+    _m_empty(); // clear MMS state
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of MMX optimized functions of class 'FIRFilter'
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "FIRFilter.h"
+
+
+FIRFilterMMX::FIRFilterMMX() : FIRFilter()
+{
+    filterCoeffsUnalign = NULL;
+}
+
+
+FIRFilterMMX::~FIRFilterMMX()
+{
+    delete[] filterCoeffsUnalign;
+}
+
+
+// (overloaded) Calculates filter coefficients for MMX routine
+void FIRFilterMMX::setCoefficients(const short *coeffs, uint newLength, uint uResultDivFactor)
+{
+    uint i;
+    FIRFilter::setCoefficients(coeffs, newLength, uResultDivFactor);
+
+    // Ensure that filter coeffs array is aligned to 16-byte boundary
+    delete[] filterCoeffsUnalign;
+    filterCoeffsUnalign = new short[2 * newLength + 8];
+    filterCoeffsAlign = (short *)SOUNDTOUCH_ALIGN_POINTER_16(filterCoeffsUnalign);
+
+    // rearrange the filter coefficients for mmx routines 
+    for (i = 0;i < length; i += 4) 
+    {
+        filterCoeffsAlign[2 * i + 0] = coeffs[i + 0];
+        filterCoeffsAlign[2 * i + 1] = coeffs[i + 2];
+        filterCoeffsAlign[2 * i + 2] = coeffs[i + 0];
+        filterCoeffsAlign[2 * i + 3] = coeffs[i + 2];
+
+        filterCoeffsAlign[2 * i + 4] = coeffs[i + 1];
+        filterCoeffsAlign[2 * i + 5] = coeffs[i + 3];
+        filterCoeffsAlign[2 * i + 6] = coeffs[i + 1];
+        filterCoeffsAlign[2 * i + 7] = coeffs[i + 3];
+    }
+}
+
+
+
+// mmx-optimized version of the filter routine for stereo sound
+uint FIRFilterMMX::evaluateFilterStereo(short *dest, const short *src, uint numSamples) const
+{
+    // Create stack copies of the needed member variables for asm routines :
+    uint i, j;
+    __m64 *pVdest = (__m64*)dest;
+
+    if (length < 2) return 0;
+
+    for (i = 0; i < (numSamples - length) / 2; i ++)
+    {
+        __m64 accu1;
+        __m64 accu2;
+        const __m64 *pVsrc = (const __m64*)src;
+        const __m64 *pVfilter = (const __m64*)filterCoeffsAlign;
+
+        accu1 = accu2 = _mm_setzero_si64();
+        for (j = 0; j < lengthDiv8 * 2; j ++)
+        {
+            __m64 temp1, temp2;
+
+            temp1 = _mm_unpacklo_pi16(pVsrc[0], pVsrc[1]);  // = l2 l0 r2 r0
+            temp2 = _mm_unpackhi_pi16(pVsrc[0], pVsrc[1]);  // = l3 l1 r3 r1
+
+            accu1 = _mm_add_pi32(accu1, _mm_madd_pi16(temp1, pVfilter[0]));  // += l2*f2+l0*f0 r2*f2+r0*f0
+            accu1 = _mm_add_pi32(accu1, _mm_madd_pi16(temp2, pVfilter[1]));  // += l3*f3+l1*f1 r3*f3+r1*f1
+
+            temp1 = _mm_unpacklo_pi16(pVsrc[1], pVsrc[2]);  // = l4 l2 r4 r2
+
+            accu2 = _mm_add_pi32(accu2, _mm_madd_pi16(temp2, pVfilter[0]));  // += l3*f2+l1*f0 r3*f2+r1*f0
+            accu2 = _mm_add_pi32(accu2, _mm_madd_pi16(temp1, pVfilter[1]));  // += l4*f3+l2*f1 r4*f3+r2*f1
+
+            // accu1 += l2*f2+l0*f0 r2*f2+r0*f0
+            //       += l3*f3+l1*f1 r3*f3+r1*f1
+
+            // accu2 += l3*f2+l1*f0 r3*f2+r1*f0
+            //          l4*f3+l2*f1 r4*f3+r2*f1
+
+            pVfilter += 2;
+            pVsrc += 2;
+        }
+        // accu >>= resultDivFactor
+        accu1 = _mm_srai_pi32(accu1, resultDivFactor);
+        accu2 = _mm_srai_pi32(accu2, resultDivFactor);
+
+        // pack 2*2*32bits => 4*16 bits
+        pVdest[0] = _mm_packs_pi32(accu1, accu2);
+        src += 4;
+        pVdest ++;
+    }
+
+   _m_empty();  // clear emms state
+
+    return (numSamples & 0xfffffffe) - length;
+}
+
+#endif  // SOUNDTOUCH_ALLOW_MMX

Externals/soundtouch/sse_optimized.cpp

@@ -0,0 +1,361 @@
+////////////////////////////////////////////////////////////////////////////////
+///
+/// SSE optimized routines for Pentium-III, Athlon-XP and later CPUs. All SSE 
+/// optimized functions have been gathered into this single source 
+/// code file, regardless to their class or original source code file, in order 
+/// to ease porting the library to other compiler and processor platforms.
+///
+/// The SSE-optimizations are programmed using SSE compiler intrinsics that
+/// are supported both by Microsoft Visual C++ and GCC compilers, so this file
+/// should compile with both toolsets.
+///
+/// NOTICE: If using Visual Studio 6.0, you'll need to install the "Visual C++ 
+/// 6.0 processor pack" update to support SSE instruction set. The update is 
+/// available for download at Microsoft Developers Network, see here:
+/// http://msdn.microsoft.com/en-us/vstudio/aa718349.aspx
+///
+/// If the above URL is expired or removed, go to "http://msdn.microsoft.com" and 
+/// perform a search with keywords "processor pack".
+///
+/// Author        : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai 'at' iki.fi
+/// SoundTouch WWW: http://www.surina.net/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed  : $Date: 2012-11-08 20:53:01 +0200 (Thu, 08 Nov 2012) $
+// File revision : $Revision: 4 $
+//
+// $Id: sse_optimized.cpp 160 2012-11-08 18:53:01Z oparviai $
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// License :
+//
+//  SoundTouch audio processing library
+//  Copyright (c) Olli Parviainen
+//
+//  This library is free software; you can redistribute it and/or
+//  modify it under the terms of the GNU Lesser General Public
+//  License as published by the Free Software Foundation; either
+//  version 2.1 of the License, or (at your option) any later version.
+//
+//  This library is distributed in the hope that it will be useful,
+//  but WITHOUT ANY WARRANTY; without even the implied warranty of
+//  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+//  Lesser General Public License for more details.
+//
+//  You should have received a copy of the GNU Lesser General Public
+//  License along with this library; if not, write to the Free Software
+//  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+//
+////////////////////////////////////////////////////////////////////////////////
+
+#include "cpu_detect.h"
+#include "STTypes.h"
+
+using namespace soundtouch;
+
+#ifdef SOUNDTOUCH_ALLOW_SSE
+
+// SSE routines available only with float sample type    
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of SSE optimized functions of class 'TDStretchSSE'
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "TDStretch.h"
+#include <xmmintrin.h>
+#include <math.h>
+
+// Calculates cross correlation of two buffers
+double TDStretchSSE::calcCrossCorr(const float *pV1, const float *pV2) const
+{
+    int i;
+    const float *pVec1;
+    const __m128 *pVec2;
+    __m128 vSum, vNorm;
+
+    // Note. It means a major slow-down if the routine needs to tolerate 
+    // unaligned __m128 memory accesses. It's way faster if we can skip 
+    // unaligned slots and use _mm_load_ps instruction instead of _mm_loadu_ps.
+    // This can mean up to ~ 10-fold difference (incl. part of which is
+    // due to skipping every second round for stereo sound though).
+    //
+    // Compile-time define SOUNDTOUCH_ALLOW_NONEXACT_SIMD_OPTIMIZATION is provided
+    // for choosing if this little cheating is allowed.
+
+#ifdef SOUNDTOUCH_ALLOW_NONEXACT_SIMD_OPTIMIZATION
+    // Little cheating allowed, return valid correlation only for 
+    // aligned locations, meaning every second round for stereo sound.
+
+    #define _MM_LOAD    _mm_load_ps
+
+    if (((ulongptr)pV1) & 15) return -1e50;    // skip unaligned locations
+
+#else
+    // No cheating allowed, use unaligned load & take the resulting
+    // performance hit.
+    #define _MM_LOAD    _mm_loadu_ps
+#endif 
+
+    // ensure overlapLength is divisible by 8
+    assert((overlapLength % 8) == 0);
+
+    // Calculates the cross-correlation value between 'pV1' and 'pV2' vectors
+    // Note: pV2 _must_ be aligned to 16-bit boundary, pV1 need not.
+    pVec1 = (const float*)pV1;
+    pVec2 = (const __m128*)pV2;
+    vSum = vNorm = _mm_setzero_ps();
+
+    // Unroll the loop by factor of 4 * 4 operations. Use same routine for
+    // stereo & mono, for mono it just means twice the amount of unrolling.
+    for (i = 0; i < channels * overlapLength / 16; i ++) 
+    {
+        __m128 vTemp;
+        // vSum += pV1[0..3] * pV2[0..3]
+        vTemp = _MM_LOAD(pVec1);
+        vSum  = _mm_add_ps(vSum,  _mm_mul_ps(vTemp ,pVec2[0]));
+        vNorm = _mm_add_ps(vNorm, _mm_mul_ps(vTemp ,vTemp));
+
+        // vSum += pV1[4..7] * pV2[4..7]
+        vTemp = _MM_LOAD(pVec1 + 4);
+        vSum  = _mm_add_ps(vSum, _mm_mul_ps(vTemp, pVec2[1]));
+        vNorm = _mm_add_ps(vNorm, _mm_mul_ps(vTemp ,vTemp));
+
+        // vSum += pV1[8..11] * pV2[8..11]
+        vTemp = _MM_LOAD(pVec1 + 8);
+        vSum  = _mm_add_ps(vSum, _mm_mul_ps(vTemp, pVec2[2]));
+        vNorm = _mm_add_ps(vNorm, _mm_mul_ps(vTemp ,vTemp));
+
+        // vSum += pV1[12..15] * pV2[12..15]
+        vTemp = _MM_LOAD(pVec1 + 12);
+        vSum  = _mm_add_ps(vSum, _mm_mul_ps(vTemp, pVec2[3]));
+        vNorm = _mm_add_ps(vNorm, _mm_mul_ps(vTemp ,vTemp));
+
+        pVec1 += 16;
+        pVec2 += 4;
+    }
+
+    // return value = vSum[0] + vSum[1] + vSum[2] + vSum[3]
+    float *pvNorm = (float*)&vNorm;
+    double norm = sqrt(pvNorm[0] + pvNorm[1] + pvNorm[2] + pvNorm[3]);
+    if (norm < 1e-9) norm = 1.0;    // to avoid div by zero
+
+    float *pvSum = (float*)&vSum;
+    return (double)(pvSum[0] + pvSum[1] + pvSum[2] + pvSum[3]) / norm;
+
+    /* This is approximately corresponding routine in C-language yet without normalization:
+    double corr, norm;
+    uint i;
+
+    // Calculates the cross-correlation value between 'pV1' and 'pV2' vectors
+    corr = norm = 0.0;
+    for (i = 0; i < channels * overlapLength / 16; i ++) 
+    {
+        corr += pV1[0] * pV2[0] +
+                pV1[1] * pV2[1] +
+                pV1[2] * pV2[2] +
+                pV1[3] * pV2[3] +
+                pV1[4] * pV2[4] +
+                pV1[5] * pV2[5] +
+                pV1[6] * pV2[6] +
+                pV1[7] * pV2[7] +
+                pV1[8] * pV2[8] +
+                pV1[9] * pV2[9] +
+                pV1[10] * pV2[10] +
+                pV1[11] * pV2[11] +
+                pV1[12] * pV2[12] +
+                pV1[13] * pV2[13] +
+                pV1[14] * pV2[14] +
+                pV1[15] * pV2[15];
+
+    for (j = 0; j < 15; j ++) norm += pV1[j] * pV1[j];
+
+        pV1 += 16;
+        pV2 += 16;
+    }
+    return corr / sqrt(norm);
+    */
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of SSE optimized functions of class 'FIRFilter'
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "FIRFilter.h"
+
+FIRFilterSSE::FIRFilterSSE() : FIRFilter()
+{
+    filterCoeffsAlign = NULL;
+    filterCoeffsUnalign = NULL;
+}
+
+
+FIRFilterSSE::~FIRFilterSSE()
+{
+    delete[] filterCoeffsUnalign;
+    filterCoeffsAlign = NULL;
+    filterCoeffsUnalign = NULL;
+}
+
+
+// (overloaded) Calculates filter coefficients for SSE routine
+void FIRFilterSSE::setCoefficients(const float *coeffs, uint newLength, uint uResultDivFactor)
+{
+    uint i;
+    float fDivider;
+
+    FIRFilter::setCoefficients(coeffs, newLength, uResultDivFactor);
+
+    // Scale the filter coefficients so that it won't be necessary to scale the filtering result
+    // also rearrange coefficients suitably for SSE
+    // Ensure that filter coeffs array is aligned to 16-byte boundary
+    delete[] filterCoeffsUnalign;
+    filterCoeffsUnalign = new float[2 * newLength + 4];
+    filterCoeffsAlign = (float *)SOUNDTOUCH_ALIGN_POINTER_16(filterCoeffsUnalign);
+
+    fDivider = (float)resultDivider;
+
+    // rearrange the filter coefficients for mmx routines 
+    for (i = 0; i < newLength; i ++)
+    {
+        filterCoeffsAlign[2 * i + 0] =
+        filterCoeffsAlign[2 * i + 1] = coeffs[i + 0] / fDivider;
+    }
+}
+
+
+
+// SSE-optimized version of the filter routine for stereo sound
+uint FIRFilterSSE::evaluateFilterStereo(float *dest, const float *source, uint numSamples) const
+{
+    int count = (int)((numSamples - length) & (uint)-2);
+    int j;
+
+    assert(count % 2 == 0);
+
+    if (count < 2) return 0;
+
+    assert(source != NULL);
+    assert(dest != NULL);
+    assert((length % 8) == 0);
+    assert(filterCoeffsAlign != NULL);
+    assert(((ulongptr)filterCoeffsAlign) % 16 == 0);
+
+    // filter is evaluated for two stereo samples with each iteration, thus use of 'j += 2'
+    for (j = 0; j < count; j += 2)
+    {
+        const float *pSrc;
+        const __m128 *pFil;
+        __m128 sum1, sum2;
+        uint i;
+
+        pSrc = (const float*)source;              // source audio data
+        pFil = (const __m128*)filterCoeffsAlign;  // filter coefficients. NOTE: Assumes coefficients 
+                                                  // are aligned to 16-byte boundary
+        sum1 = sum2 = _mm_setzero_ps();
+
+        for (i = 0; i < length / 8; i ++) 
+        {
+            // Unroll loop for efficiency & calculate filter for 2*2 stereo samples 
+            // at each pass
+
+            // sum1 is accu for 2*2 filtered stereo sound data at the primary sound data offset
+            // sum2 is accu for 2*2 filtered stereo sound data for the next sound sample offset.
+
+            sum1 = _mm_add_ps(sum1, _mm_mul_ps(_mm_loadu_ps(pSrc)    , pFil[0]));
+            sum2 = _mm_add_ps(sum2, _mm_mul_ps(_mm_loadu_ps(pSrc + 2), pFil[0]));
+
+            sum1 = _mm_add_ps(sum1, _mm_mul_ps(_mm_loadu_ps(pSrc + 4), pFil[1]));
+            sum2 = _mm_add_ps(sum2, _mm_mul_ps(_mm_loadu_ps(pSrc + 6), pFil[1]));
+
+            sum1 = _mm_add_ps(sum1, _mm_mul_ps(_mm_loadu_ps(pSrc + 8) ,  pFil[2]));
+            sum2 = _mm_add_ps(sum2, _mm_mul_ps(_mm_loadu_ps(pSrc + 10), pFil[2]));
+
+            sum1 = _mm_add_ps(sum1, _mm_mul_ps(_mm_loadu_ps(pSrc + 12), pFil[3]));
+            sum2 = _mm_add_ps(sum2, _mm_mul_ps(_mm_loadu_ps(pSrc + 14), pFil[3]));
+
+            pSrc += 16;
+            pFil += 4;
+        }
+
+        // Now sum1 and sum2 both have a filtered 2-channel sample each, but we still need
+        // to sum the two hi- and lo-floats of these registers together.
+
+        // post-shuffle & add the filtered values and store to dest.
+        _mm_storeu_ps(dest, _mm_add_ps(
+                    _mm_shuffle_ps(sum1, sum2, _MM_SHUFFLE(1,0,3,2)),   // s2_1 s2_0 s1_3 s1_2
+                    _mm_shuffle_ps(sum1, sum2, _MM_SHUFFLE(3,2,1,0))    // s2_3 s2_2 s1_1 s1_0
+                    ));
+        source += 4;
+        dest += 4;
+    }
+
+    // Ideas for further improvement:
+    // 1. If it could be guaranteed that 'source' were always aligned to 16-byte 
+    //    boundary, a faster aligned '_mm_load_ps' instruction could be used.
+    // 2. If it could be guaranteed that 'dest' were always aligned to 16-byte 
+    //    boundary, a faster '_mm_store_ps' instruction could be used.
+
+    return (uint)count;
+
+    /* original routine in C-language. please notice the C-version has differently 
+       organized coefficients though.
+    double suml1, suml2;
+    double sumr1, sumr2;
+    uint i, j;
+
+    for (j = 0; j < count; j += 2)
+    {
+        const float *ptr;
+        const float *pFil;
+
+        suml1 = sumr1 = 0.0;
+        suml2 = sumr2 = 0.0;
+        ptr = src;
+        pFil = filterCoeffs;
+        for (i = 0; i < lengthLocal; i ++) 
+        {
+            // unroll loop for efficiency.
+
+            suml1 += ptr[0] * pFil[0] + 
+                     ptr[2] * pFil[2] +
+                     ptr[4] * pFil[4] +
+                     ptr[6] * pFil[6];
+
+            sumr1 += ptr[1] * pFil[1] + 
+                     ptr[3] * pFil[3] +
+                     ptr[5] * pFil[5] +
+                     ptr[7] * pFil[7];
+
+            suml2 += ptr[8] * pFil[0] + 
+                     ptr[10] * pFil[2] +
+                     ptr[12] * pFil[4] +
+                     ptr[14] * pFil[6];
+
+            sumr2 += ptr[9] * pFil[1] + 
+                     ptr[11] * pFil[3] +
+                     ptr[13] * pFil[5] +
+                     ptr[15] * pFil[7];
+
+            ptr += 16;
+            pFil += 8;
+        }
+        dest[0] = (float)suml1;
+        dest[1] = (float)sumr1;
+        dest[2] = (float)suml2;
+        dest[3] = (float)sumr2;
+
+        src += 4;
+        dest += 4;
+    }
+    */
+}
+
+#endif  // SOUNDTOUCH_ALLOW_SSE