comparison toolboxes/MIRtoolbox1.3.2/MIRToolbox/mirflux.m @ 0:e9a9cd732c1e tip

first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:e9a9cd732c1e
1 function varargout = mirflux(orig,varargin)
2 % f = mirflux(x) measures distance between successive frames.
3 % First argument:
4 % If x is a spectrum, this corresponds to spectral flux.
5 % But the flux of any other data can be computed as well.
6 % If x is an audio file or audio signal, the spectral flux is
7 % computed by default.
8 % Optional arguments:
9 % f = mirflux(x,'Frame',...) specifies the frame parameters, if x is
10 % not already decomposed into frames. Default values are frame
11 % length of .2 seconds and hop factor of 1.3.
12 % f = mirflux(x,'Dist',d) specifies the distance between
13 % successive frames: (IF 'COMPLEX': DISTANCE = 'CITY' ALWAYS)
14 % d = 'Euclidian': Euclidian distance (Default)
15 % d = 'City': City-block distance
16 % d = 'Cosine': Cosine distance (or normalized correlation)
17 % f = mirflux(...,'Inc'): Only positive difference between frames are
18 % summed, in order to focus on increase of energy solely.
19 % f = mirflux(...,'Complex'), for spectral flux, combines use of
20 % energy and phase information (Bello et al, 2004).
21 % f = mirflux(...,'Halfwave'): performs a half-wave rectification on
22 % the result.
23 % f = mirflux(...,'Median',l,C): removes small spurious peaks by
24 % subtracting to the result its median filtering. The median
25 % filter computes the point-wise median inside a window of length
26 % l (in seconds), that includes a same number of previous and
27 % next samples. C is a scaling factor whose purpose is to
28 % artificially rise the curve slightly above the steady state of
29 % the signal. If no parameters are given, the default values are:
30 % l = 0.2 s. and C = 1.3
31 % f = mirflux(...,'Median',l,C,'Halfwave'): The scaled median
32 % filtering is designed to be succeeded by the half-wave
33 % rectification process in order to select peaks above the
34 % dynamic threshold calculated with the help of the median
35 % filter. The resulting signal is called "detection function"
36 % (Alonso, David, Richard, 2004). To ensure accurate detection,
37 % the length l of the median filter must be longer than the
38 % average width of the peaks of the detection function.
39 %
40 % (Bello et al, 2004) Juan P. Bello, Chris Duxbury, Mike Davies, and Mark
41 % Sandler, On the Use of Phase and Energy for Musical Onset Detection in
42 % the Complex Domain, IEEE SIGNAL PROCESSING LETTERS, VOL. 11, NO. 6,
43 % JUNE 2004
44
45 dist.key = 'Dist';
46 dist.type = 'String';
47 dist.choice = {'Euclidian','City','Cosine'}; %PDIST???? (euclidean)
48 dist.default = 'Euclidian';
49 option.dist = dist;
50
51 inc.key = 'Inc';
52 inc.type = 'Boolean';
53 inc.default = 0;
54 option.inc = inc;
55
56 complex.key = 'Complex';
57 complex.type = 'Boolean';
58 complex.default = 0;
59 option.complex = complex;
60
61 h.key = 'Halfwave';
62 h.type = 'Boolean';
63 h.default = 0;
64 h.when = 'After';
65 option.h = h;
66
67 median.key = 'Median';
68 median.type = 'Integer';
69 median.number = 2;
70 median.default = [0 0];
71 median.keydefault = [.2 1.3];
72 median.when = 'After';
73 option.median = median;
74
75 frame.key = 'Frame';
76 frame.type = 'Integer';
77 frame.number = 2;
78 frame.default = [.05 .5];
79 option.frame = frame;
80
81 specif.option = option;
82
83 varargout = mirfunction(@mirflux,orig,varargin,nargout,specif,@init,@main);
84
85
86 function [x type] = init(x,option)
87 if isamir(x,'miraudio')
88 if isframed(x)
89 x = mirspectrum(x);
90 else
91 x = mirspectrum(x,'Frame',option.frame.length.val,option.frame.length.unit,...
92 option.frame.hop.val,option.frame.hop.unit);
93 end
94 end
95 if isa(x,'mirdesign')
96 x = set(x,'Overlap',1);
97 end
98 type = 'mirscalar';
99
100
101 function f = main(s,option,postoption)
102 if iscell(s)
103 s = s{1};
104 end
105 t = get(s,'Title');
106 if isa(s,'mirscalar') && ...
107 (strcmp(t,'Harmonic Change Detection Function') || ...
108 (length(t)>4 && strcmp(t(end-3:end),'flux')) || ...
109 (length(t)>5 && strcmp(t(end-4:end-1),'flux')))
110 if not(isempty(postoption))
111 f = modif(s,postoption);
112 end
113 else
114 if isa(s,'mirspectrum')
115 t = 'Spectral';
116 end
117 m = get(s,'Data');
118 if option.complex
119 if not(isa(s,'mirspectrum'))
120 error('ERROR IN MIRFLUX: ''Complex'' option only defined for spectral flux.');
121 end
122 ph = get(s,'Phase');
123 end
124 param.complex = option.complex;
125 param.inc = option.inc;
126 fp = get(s,'FramePos');
127 if strcmp(t,'Tonal centroid')
128 t = 'Harmonic Change Detection Function';
129 else
130 t = [t,' flux'];
131 end
132 disp(['Computing ' t '...'])
133 ff = cell(1,length(m));
134 newsr = cell(1,length(m));
135 dist = str2func(option.dist);
136 %[tmp s] = gettmp(s); %get(s,'Tmp');
137 for h = 1:length(m)
138 ff{h} = cell(1,length(m{h}));
139 if not(iscell(m{h}))
140 m{h} = {m{h}};
141 end
142 for i = 1:length(m{h})
143 mi = m{h}{i};
144 if size(mi,3) > 1 && size(mi,1) == 1
145 mi = reshape(mi,size(mi,2),size(mi,3))';
146 end
147 if option.complex
148 phi = ph{h}{i};
149 end
150 fpi = fp{h}{i};
151 %if 0 %not(isempty(tmp)) && isstruct(tmp) && isfield(tmp,'mi')
152 % mi = [tmp.mi mi];
153 % fpi = [tmp.fpi fpi];
154 %end
155 nc = size(mi,2);
156 np = size(mi,3);
157 if nc == 1
158 warning('WARNING IN MIRFLUX: Flux can only be computed on signal decomposed into frames.');
159 ff{h}{i} = NaN(1,1,np);
160 else
161 if option.complex
162 fl = zeros(1,nc-2,np);
163 for k = 1:np
164 d = diff(phi(:,:,k),2,2);
165 d = d/(2*pi) - round(d/(2*pi));
166 g = sqrt(mi(:,3:end,k).^2 + mi(:,2:(end-1),k).^2 ...
167 - 2.*mi(:,3:end,k)...
168 .*mi(:,2:(end-1),k)...
169 .*cos(d));
170 fl(1,:,k) = sum(g);
171 end
172 fp{h}{i} = fpi(:,3:end);
173 else
174 fl = zeros(1,nc-1,np);
175 for k = 1:np
176 for j = 1:nc-1
177 fl(1,j,k) = dist(mi(:,j,k),mi(:,j+1,k),option.inc);
178 end
179 end
180 fp{h}{i} = fpi(:,2:end);
181 end
182 ff{h}{i} = fl;
183 %tmp.mi = mi(:,end,:);
184 %tmp.fpi = fpi(:,end,:);
185 end
186 end
187 %tmp = [];
188 if size(fpi,2)<2
189 newsr{h} = 0;
190 else
191 newsr{h} = 1/(fpi(1,2)-fpi(1,1));
192 end
193 end
194 f = mirscalar(s,'Data',ff,'FramePos',fp,'Sampling',newsr,...
195 'Title',t,'Parameter',param); %,'Tmp',tmp);
196 %f = settmp(f,tmp);
197 if not(isempty(postoption))
198 f = modif(f,postoption);
199 end
200 end
201
202
203 function f = modif(f,option)
204 fl = get(f,'Data');
205 r = get(f,'Sampling');
206 for h = 1:length(fl)
207 for i = 1:length(fl{h})
208 fli = fl{h}{i};
209 nc = size(fli,2);
210 np = size(fli,3);
211 if option.median(1)
212 ffi = zeros(1,nc,np);
213 lr = round(option.median(1)*r{i});
214 for k = 1:np
215 for j = 1:nc
216 ffi(:,j,k) = fli(:,j,k) - ...
217 option.median(2) * median(fli(:,max(1,j-lr):min(nc-1,j+lr),k));
218 end
219 end
220 fli = ffi;
221 end
222 if option.h
223 fli = hwr(fli);
224 end
225 fl{h}{i} = fli;
226 end
227 end
228 f = set(f,'Data',fl);
229
230
231 function y = Euclidian(mi,mj,inc)
232 if inc
233 y = sqrt(sum(max(0,(mj-mi)).^2));
234 else
235 y = sqrt(sum((mj-mi).^2));
236 end
237
238
239 function y = City(mi,mj,inc)
240 if inc
241 y = sum(max(0,(mj-mi)));
242 else
243 y = sum(abs(mj-mi));
244 end
245
246
247 function d = Cosine(r,s,inc)
248 nr = sqrt(r'*r);
249 ns = sqrt(s'*s);
250 if or(nr == 0, ns == 0);
251 d = 1;
252 else
253 d = 1 - r'*s/nr/ns;
254 end