c@45: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
c@45: 
c@45: /*
c@45:     QM Vamp Plugin Set
c@45: 
c@45:     Centre for Digital Music, Queen Mary, University of London.
c@135: 
c@135:     This program is free software; you can redistribute it and/or
c@135:     modify it under the terms of the GNU General Public License as
c@135:     published by the Free Software Foundation; either version 2 of the
c@135:     License, or (at your option) any later version.  See the file
c@135:     COPYING included with this distribution for more information.
c@45: */
c@45: 
c@45: #include "MFCCPlugin.h"
c@45: 
c@45: #include <dsp/mfcc/MFCC.h>
c@130: #include <maths/MathUtilities.h>
c@54: 
c@45: using std::string;
c@45: using std::vector;
c@45: using std::cerr;
c@45: using std::endl;
c@45: 
c@45: MFCCPlugin::MFCCPlugin(float inputSampleRate) :
c@45:     Vamp::Plugin(inputSampleRate),
c@45:     m_config(lrintf(inputSampleRate)),
c@45:     m_mfcc(0),
c@45:     m_step(1024),
c@60:     m_block(2048),
c@60:     m_count(0)
c@45: {
c@45:     m_bins = 20;
c@45:     m_wantC0 = true;
c@45:     m_logpower = 1;
c@45: 
c@45:     setupConfig();
c@45: }
c@45: 
c@45: void
c@45: MFCCPlugin::setupConfig()
c@45: {
c@45:     m_config.FS = lrintf(m_inputSampleRate);
c@45:     m_config.fftsize = m_block;
c@45:     m_config.nceps = (m_wantC0 ? m_bins-1 : m_bins);
c@45:     m_config.want_c0 = m_wantC0;
c@45:     m_config.logpower = m_logpower;
c@45: }
c@45: 
c@45: MFCCPlugin::~MFCCPlugin()
c@45: {
c@45:     delete m_mfcc;
c@45: }
c@45: 
c@45: string
c@45: MFCCPlugin::getIdentifier() const
c@45: {
c@45:     return "qm-mfcc";
c@45: }
c@45: 
c@45: string
c@45: MFCCPlugin::getName() const
c@45: {
c@45:     return "Mel-Frequency Cepstral Coefficients";
c@45: }
c@45: 
c@45: string
c@45: MFCCPlugin::getDescription() const
c@45: {
c@50:     return "Calculate a series of MFCC vectors from the audio";
c@45: }
c@45: 
c@45: string
c@45: MFCCPlugin::getMaker() const
c@45: {
c@45:     return "Queen Mary, University of London";
c@45: }
c@45: 
c@45: int
c@45: MFCCPlugin::getPluginVersion() const
c@45: {
c@45:     return 1;
c@45: }
c@45: 
c@45: string
c@45: MFCCPlugin::getCopyright() const
c@45: {
c@118:     return "Plugin by Nicolas Chetry and Chris Cannam.  Copyright (c) 2009 QMUL - All Rights Reserved";
c@45: }
c@45: 
c@45: MFCCPlugin::ParameterList
c@45: MFCCPlugin::getParameterDescriptors() const
c@45: {
c@45:     ParameterList list;
c@45: 
c@45:     ParameterDescriptor desc;
c@45:     desc.identifier = "nceps";
c@45:     desc.name = "Number of Coefficients";
c@45:     desc.unit = "";
c@52:     desc.description = "Number of MFCCs to return, starting from C0 if \"Include C0\" is specified or from C1 otherwise";
c@45:     desc.minValue = 1;
c@45:     desc.maxValue = 40;
c@45:     desc.defaultValue = 20;
c@45:     desc.isQuantized = true;
c@45:     desc.quantizeStep = 1;
c@45:     list.push_back(desc);
c@45: 
c@45:     desc.identifier = "logpower";
c@45:     desc.name = "Power for Mel Amplitude Logs";
c@45:     desc.unit = "";
c@52:     desc.description = "Power to raise the amplitude log values to before applying DCT.  Values greater than 1 may reduce contribution of noise";
c@45:     desc.minValue = 0;
c@45:     desc.maxValue = 5;
c@45:     desc.defaultValue = 1;
c@45:     desc.isQuantized = false;
c@45:     desc.quantizeStep = 0;
c@45:     list.push_back(desc);
c@45: 
c@45:     desc.identifier = "wantc0";
c@45:     desc.name = "Include C0";
c@45:     desc.unit = "";
c@52:     desc.description = "Whether to include the C0 (energy level) coefficient in the returned results";
c@45:     desc.minValue = 0;
c@45:     desc.maxValue = 1;
c@45:     desc.defaultValue = 1;
c@45:     desc.isQuantized = true;
c@45:     desc.quantizeStep = 1;
c@45:     list.push_back(desc);
c@45: 
c@45:     return list;
c@45: }
c@45: 
c@45: float
c@45: MFCCPlugin::getParameter(std::string param) const
c@45: {
c@45:     if (param == "nceps") {
c@45:         return m_bins;
c@45:     }
c@45:     if (param == "logpower") {
c@45:         return m_logpower;
c@45:     }
c@45:     if (param == "wantc0") {
c@45:         return m_wantC0 ? 1 : 0;
c@45:     }
c@45:     std::cerr << "WARNING: MFCCPlugin::getParameter: unknown parameter \""
c@45:               << param << "\"" << std::endl;
c@45:     return 0.0;
c@45: }
c@45: 
c@45: void
c@45: MFCCPlugin::setParameter(std::string param, float value)
c@45: {
c@45:     if (param == "nceps") {
c@45:         m_bins = lrintf(value);
c@45:     } else if (param == "logpower") {
c@45:         m_logpower = lrintf(value);
c@45:     } else if (param == "wantc0") {
c@45:         m_wantC0 = (value > 0.5);
c@45:     } else {
c@45:         std::cerr << "WARNING: MFCCPlugin::setParameter: unknown parameter \""
c@45:                   << param << "\"" << std::endl;
c@45:     }
c@45: 
c@45:     setupConfig();
c@45: }
c@45: 
c@45: bool
c@45: MFCCPlugin::initialise(size_t channels, size_t stepSize, size_t blockSize)
c@45: {
c@45:     if (m_mfcc) {
c@45: 	delete m_mfcc;
c@45: 	m_mfcc = 0;
c@45:     }
c@45: 
c@45:     if (channels < getMinChannelCount() ||
c@45: 	channels > getMaxChannelCount()) return false;
c@45: 
c@95: //    std::cerr << "MFCCPlugin::initialise: step " << stepSize << ", block "
c@95: //	      << blockSize << std::endl;
c@45: 
c@45:     m_step = stepSize;
c@45:     m_block = blockSize;
c@45:     setupConfig();
c@45: 
c@45:     m_mfcc = new MFCC(m_config);
c@45: 
c@45:     m_binsums = vector<double>(m_bins);
c@45:     for (int i = 0; i < m_bins; ++i) {
c@45:         m_binsums[i] = 0.0;
c@45:     }
c@45: 
c@45:     return true;
c@45: }
c@45: 
c@45: void
c@45: MFCCPlugin::reset()
c@45: {
c@45:     if (m_mfcc) {
c@45: 	delete m_mfcc;
c@45: 	m_mfcc = new MFCC(m_config);
c@45:         for (int i = 0; i < m_bins; ++i) {
c@45:             m_binsums[i] = 0.0;
c@45:         }
c@45:     }
c@60:     m_count = 0;
c@45: }
c@45: 
c@45: size_t
c@45: MFCCPlugin::getPreferredStepSize() const
c@45: {
c@45:     return 1024;
c@45: }
c@45: 
c@45: size_t
c@45: MFCCPlugin::getPreferredBlockSize() const
c@45: {
c@45:     return 2048;
c@45: }
c@45: 
c@45: MFCCPlugin::OutputList
c@45: MFCCPlugin::getOutputDescriptors() const
c@45: {
c@45:     OutputList list;
c@45: 
c@45:     OutputDescriptor d;
c@45:     d.identifier = "coefficients";
c@45:     d.name = "Coefficients";
c@45:     d.unit = "";
c@52:     d.description = "MFCC values";
c@45:     d.hasFixedBinCount = true;
c@45:     d.binCount = m_bins;
c@45:     d.hasKnownExtents = false;
c@45:     d.isQuantized = false;
c@45:     d.sampleType = OutputDescriptor::OneSamplePerStep;
c@45:     list.push_back(d);
c@45: 
c@45:     d.identifier = "means";
c@45:     d.name = "Means of Coefficients";
c@52:     d.description = "Mean values of MFCCs across duration of audio input";
c@45:     d.sampleType = OutputDescriptor::FixedSampleRate;
c@45:     d.sampleRate = 1;
c@45:     list.push_back(d);
c@45: 
c@45:     return list;
c@45: }
c@45: 
c@45: MFCCPlugin::FeatureSet
c@45: MFCCPlugin::process(const float *const *inputBuffers,
c@45:                     Vamp::RealTime /* timestamp */)
c@45: {
c@45:     if (!m_mfcc) {
c@45: 	cerr << "ERROR: MFCCPlugin::process: "
c@45: 	     << "MFCC has not been initialised"
c@45: 	     << endl;
c@45: 	return FeatureSet();
c@45:     }
c@45: 
c@45:     double *real = new double[m_block];
c@45:     double *imag = new double[m_block];
c@45: 
c@75:     for (size_t i = 0; i <= m_block/2; ++i) {
c@45: 	real[i] = inputBuffers[0][i*2];
c@45: 	if (i > 0) real[m_block - i] = real[i];
c@45:         imag[i] = inputBuffers[0][i*2+1];
c@45:         if (i > 0) imag[m_block - i] = imag[i];
c@45:     }
c@45: 
c@45:     double *output = new double[m_bins];
c@45: 
c@45:     m_mfcc->process(real, imag, output);
c@45: 
c@45:     delete[] real;
c@45:     delete[] imag;
c@45: 
c@45:     Feature feature;
c@45:     feature.hasTimestamp = false;
c@178:     for (int i = 0; i < m_bins; ++i) {
c@45:         double value = output[i];
c@130:         if (ISNAN(value)) value = 0.0;
c@45:         m_binsums[i] += value;
c@45: 	feature.values.push_back(value);
c@45:     }
c@45:     feature.label = "";
c@45:     ++m_count;
c@45: 
c@95:     delete[] output;
c@95: 
c@45:     FeatureSet returnFeatures;
c@45:     returnFeatures[0].push_back(feature);
c@45:     return returnFeatures;
c@45: }
c@45: 
c@45: MFCCPlugin::FeatureSet
c@45: MFCCPlugin::getRemainingFeatures()
c@45: {
c@45:     Feature feature;
c@45:     feature.hasTimestamp = true;
c@45:     feature.timestamp = Vamp::RealTime::zeroTime;
c@45:   
c@178:     for (int i = 0; i < m_bins; ++i) {
c@45:         double v = m_binsums[i];
c@45:         if (m_count > 0) v /= m_count;
c@45:         feature.values.push_back(v);
c@45:     }
c@45:     feature.label = "Coefficient means";
c@45: 
c@45:     FeatureSet returnFeatures;
c@45:     returnFeatures[1].push_back(feature);
c@45:     return returnFeatures;
c@45: }
c@45: