c@243
|
1 /*
|
c@243
|
2 * ClusterMeltSegmenter.h
|
c@243
|
3 * soundbite
|
c@243
|
4 *
|
c@243
|
5 * Created by Mark Levy on 23/03/2006.
|
c@243
|
6 * Copyright 2006 Centre for Digital Music, Queen Mary, University of London. All rights reserved.
|
c@243
|
7 *
|
c@243
|
8 */
|
c@243
|
9
|
c@243
|
10 #include <vector>
|
c@243
|
11
|
c@243
|
12 #include "segment.h"
|
c@243
|
13 #include "Segmenter.h"
|
c@243
|
14 #include "hmm.h"
|
c@243
|
15 #include "lib_constQ.h"
|
c@243
|
16
|
c@243
|
17 using std::vector;
|
c@243
|
18
|
c@243
|
19 class ClusterMeltSegmenterParams // defaults are sensible for 11025Hz with 0.2 second hopsize
|
c@243
|
20 {
|
c@243
|
21 public:
|
c@243
|
22 ClusterMeltSegmenterParams() : featureType(FEATURE_TYPE_CONSTQ), hopSize(0.2), windowSize(0.6), fmin(62), fmax(16000),
|
c@243
|
23 nbins(8), ncomponents(20), nHMMStates(40), nclusters(10), histogramLength(15), neighbourhoodLimit(20) { }
|
c@243
|
24 feature_types featureType;
|
c@243
|
25 double hopSize; // in secs
|
c@243
|
26 double windowSize; // in secs
|
c@243
|
27 int fmin;
|
c@243
|
28 int fmax;
|
c@243
|
29 int nbins;
|
c@243
|
30 int ncomponents;
|
c@243
|
31 int nHMMStates;
|
c@243
|
32 int nclusters;
|
c@243
|
33 int histogramLength;
|
c@243
|
34 int neighbourhoodLimit;
|
c@243
|
35 };
|
c@243
|
36
|
c@243
|
37 class ClusterMeltSegmenter : public Segmenter
|
c@243
|
38 {
|
c@243
|
39 public:
|
c@243
|
40 ClusterMeltSegmenter(ClusterMeltSegmenterParams params);
|
c@243
|
41 virtual ~ClusterMeltSegmenter();
|
c@243
|
42 virtual void initialise(int samplerate);
|
c@243
|
43 virtual int getWindowsize() { return static_cast<int>(windowSize * samplerate); }
|
c@243
|
44 virtual int getHopsize() { return static_cast<int>(hopSize * samplerate); }
|
c@243
|
45 virtual void extractFeatures(double* samples, int nsamples);
|
c@243
|
46 void setFeatures(const vector<vector<double> >& f); // provide the features yourself
|
c@243
|
47 virtual void segment(); // segment into default number of segment-types
|
c@243
|
48 void segment(int m); // segment into m segment-types
|
c@243
|
49 int getNSegmentTypes() { return nclusters; }
|
c@243
|
50 protected:
|
c@243
|
51 //void mpeg7ConstQ();
|
c@243
|
52 void makeSegmentation(int* q, int len);
|
c@243
|
53
|
c@243
|
54 double* window;
|
c@243
|
55 int windowLength; // in samples
|
c@243
|
56 constQ_t* constq;
|
c@243
|
57 model_t* model; // the HMM
|
c@243
|
58 //vector<int> stateSequence;
|
c@243
|
59 //vector<int> segmentTypeSequence;
|
c@243
|
60 int* q; // the decoded HMM state sequence
|
c@243
|
61 vector<vector<double> > histograms;
|
c@243
|
62
|
c@243
|
63 feature_types featureType;
|
c@243
|
64 double hopSize; // in seconds
|
c@243
|
65 double windowSize; // in seconds
|
c@243
|
66
|
c@243
|
67 // constant-Q parameters
|
c@243
|
68 int fmin;
|
c@243
|
69 int fmax;
|
c@243
|
70 int nbins;
|
c@243
|
71 int ncoeff;
|
c@243
|
72
|
c@243
|
73 // PCA parameters
|
c@243
|
74 int ncomponents;
|
c@243
|
75
|
c@243
|
76 // HMM parameters
|
c@243
|
77 int nHMMStates;
|
c@243
|
78
|
c@243
|
79 // clustering parameters
|
c@243
|
80 int nclusters;
|
c@243
|
81 int histogramLength;
|
c@243
|
82 int neighbourhoodLimit;
|
c@243
|
83 };
|