SoundFeatureToolbox/processclusters_fn.m at master · lestheprof/SoundFeatureToolbox · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
function [fem, male] = processclustersica_fn(timitlocn, vdirectory, segmentsfile, clustersfile , varargin)
% processclusters_fn: using the segments in the segmentsfile, and the
% clusters in the clustersfile, find and return the phonemes (and their
% frequencies) in each cluster

% LSS 10 March 2014.
% updated 2 April 2014, ICA only (PCA is different)
%
const_phonfraction = 0.5 ; % fraction of phonemesin those returned.
i=1 ;
while(i<=size(varargin,2))
    switch lower(varargin{i})
        case 'const_phonfraction';
            const_phonfraction=varargin{i+1};
            i=i+1;

        otherwise
            error('processclusters_fn: Unknown argument %s given',varargin{i});
    end
    i=i+1;
end
%
% read the segmentsfiles
% we only need f1 and m1 from them
segdata = load([timitlocn '/' vdirectory '/' segmentsfile]) ;
% read the clustersfile
clustdata_m = load([timitlocn '/' vdirectory '/' clustersfile '_m']) ;
clustdata_f = load([timitlocn '/' vdirectory '/' clustersfile '_f']) ;

% female first
%
% deal with PCA first


% and ICAs

for  i = 1:clustdata_f.icaparams.clusters_ica % for each cluster
    icaphonemes{i} = segdata.f1.segmentphoneme(clustdata_f.line1to8_km5_ica == i) ; % find the phonemes in this cluster
    % find the list of phonemes, and the index in that list of each
    [icaphonemesunique{i}, ~, icaphuid{i}] = unique(icaphonemes{i}) ;

    % now find the most frequently occurring phonemes in each of the
    % n_clusters_pca lists.
    numphonemes = length(icaphonemes{i}) ;
    % intiialise arrays
    modeno = zeros([1 floor(numphonemes/2)]) ;
    phonfreq = zeros([1 floor(numphonemes/2)]) ;
    totalfrequency = 0 ;
    phonsinlist = 0 ;
    icaphuidtemp = icaphuid{i} ;
    while (totalfrequency < (numphonemes * const_phonfraction))
        phonsinlist = phonsinlist + 1 ;
        [modeno(phonsinlist), phonfreq(phonsinlist)] = mode(icaphuidtemp) ;
        icaphuidtemp= icaphuidtemp(icaphuidtemp ~= modeno(phonsinlist)) ;
        totalfrequency = totalfrequency + phonfreq(phonsinlist) ;
    end
    %modeno now has the most common phonemes indexes, and phonfreq has the
    %frequency of each. Now put the actual phoneme into an array
    fem.icamodephonemes{i} = icaphonemesunique{i}(modeno(1:phonsinlist)) ;
    fem.icaphonfreqclust{i} = phonfreq(1:phonsinlist) ;
    fem.icaphons(i) = numphonemes ;

end
% male next
%

% and ICAs
for  i = 1:clustdata_m.icaparams.clusters_ica % for each cluster
    icaphonemes{i} = segdata.m1.segmentphoneme(clustdata_m.line1to8_km5_ica == i) ; % find the phonemes in this cluster
    % find the list of phonemes, and the index in that list of each
    [icaphonemesunique{i}, ~, icaphuid{i}] = unique(icaphonemes{i}) ;

    % now find the most frequently occurring phonemes in each of the
    % n_clusters_pca lists.
    numphonemes = length(icaphonemes{i}) ;
    % intiialise arrays
    modeno = zeros([1 floor(numphonemes/2)]) ;
    phonfreq = zeros([1 floor(numphonemes/2)]) ;
    totalfrequency = 0 ;
    phonsinlist = 0 ;
    icaphuidtemp = icaphuid{i} ;
    while (totalfrequency < (numphonemes * const_phonfraction))
        phonsinlist = phonsinlist + 1 ;
        [modeno(phonsinlist), phonfreq(phonsinlist)] = mode(icaphuidtemp) ;
        icaphuidtemp= icaphuidtemp(icaphuidtemp ~= modeno(phonsinlist)) ;
        totalfrequency = totalfrequency + phonfreq(phonsinlist) ;
    end
    %modeno now has the most common phonemes indexes, and phonfreq has the
    %frequency of each. Now put the actual phoneme into an array
    male.icamodephonemes{i} = icaphonemesunique{i}(modeno(1:phonsinlist)) ;
    male.icaphonfreqclust{i} = phonfreq(1:phonsinlist) ;
    male.icaphons = numphonemes ;

end
end