c@249: /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
c@249: 
c@243: /*
c@249:  * ClusterMeltSegmenter.h
c@243:  *
c@249:  * Created by Mark Levy on 23/03/2006.
c@249:  * Copyright 2006 Centre for Digital Music, Queen Mary, University of London.
c@309: 
c@309:     This program is free software; you can redistribute it and/or
c@309:     modify it under the terms of the GNU General Public License as
c@309:     published by the Free Software Foundation; either version 2 of the
c@309:     License, or (at your option) any later version.  See the file
c@309:     COPYING included with this distribution for more information.
c@243:  */
c@243: 
cannam@491: #ifndef QM_DSP_CLUSTER_MELT_SEGMENTER_H
cannam@491: #define QM_DSP_CLUSTER_MELT_SEGMENTER_H
cannam@491: 
c@243: #include <vector>
c@243: 
c@243: #include "segment.h"
c@243: #include "Segmenter.h"
c@245: #include "hmm/hmm.h"
c@245: #include "base/Window.h"
c@243: 
c@249: class Decimator;
c@249: class ConstantQ;
c@251: class MFCC;
c@289: class FFTReal;
c@249: 
c@249: class ClusterMeltSegmenterParams
c@249: // defaults are sensible for 11025Hz with 0.2 second hopsize
c@243: {
c@243: public:
c@249:     ClusterMeltSegmenterParams() : 
c@249:         featureType(FEATURE_TYPE_CONSTQ),
c@249:         hopSize(0.2),
c@249:         windowSize(0.6),
c@249:         fmin(62),
c@249:         fmax(16000), 
c@249:         nbins(8),
c@249:         ncomponents(20),
cannam@480:         nHMMStates(40),
c@249:         nclusters(10),
c@249:         histogramLength(15),
c@249:         neighbourhoodLimit(20) { }
c@249:     feature_types featureType;
cannam@480:     double hopSize;     // in secs
cannam@480:     double windowSize;  // in secs
c@249:     int fmin;
c@249:     int fmax;
c@249:     int nbins;
c@249:     int ncomponents;
c@249:     int nHMMStates;
c@249:     int nclusters;
c@249:     int histogramLength;
c@249:     int neighbourhoodLimit;
c@243: };
c@243: 
c@243: class ClusterMeltSegmenter : public Segmenter
c@243: {
c@243: public:
c@249:     ClusterMeltSegmenter(ClusterMeltSegmenterParams params);
c@249:     virtual ~ClusterMeltSegmenter();
c@249:     virtual void initialise(int samplerate);
c@249:     virtual int getWindowsize();
c@249:     virtual int getHopsize();
c@249:     virtual void extractFeatures(const double* samples, int nsamples);
cannam@493:     void setFeatures(const std::vector<std::vector<double> >& f);         // provide the features yourself
cannam@480:     virtual void segment();             // segment into default number of segment-types
cannam@480:     void segment(int m);                // segment into m segment-types
c@249:     int getNSegmentTypes() { return nclusters; }
c@249: 
c@243: protected:
c@249:     void makeSegmentation(int* q, int len);
cannam@480:         
c@251:     void extractFeaturesConstQ(const double *, int);
c@251:     void extractFeaturesMFCC(const double *, int);
c@251: 
c@249:     Window<double> *window;
c@289:     FFTReal *fft;
c@251:     ConstantQ* constq; 
c@251:     MFCC* mfcc;
cannam@480:     model_t* model;                             // the HMM
cannam@480:     int* q;                                     // the decoded HMM state sequence
cannam@493:     std::vector<std::vector<double> > histograms; 
c@249:     
cannam@480:     feature_types featureType;  
cannam@480:     double hopSize;             // in seconds
cannam@480:     double windowSize;  // in seconds
c@249:     
c@249:     // constant-Q parameters
c@249:     int fmin;
c@249:     int fmax;
c@249:     int nbins;
c@249:     int ncoeff;
c@249:     
c@249:     // PCA parameters
c@249:     int ncomponents;
c@249:     
c@249:     // HMM parameters
c@249:     int nHMMStates;
c@249:     
c@249:     // clustering parameters
c@249:     int nclusters;
c@249:     int histogramLength;
c@249:     int neighbourhoodLimit;
c@249: 
c@249:     Decimator *decimator;
c@243: };
cannam@491: 
cannam@491: #endif