wolffd@0: function demo4segmentation
wolffd@0: % To get familiar with some approaches of segmentation of audio files 
wolffd@0: % using MIRtoolbox. 
wolffd@0:  
wolffd@0: % 1. Load an audio file (for instance, guitar.wav). 
wolffd@0: a = miraudio('guitar');
wolffd@0: 
wolffd@0: % 2. We will perform the segmentation strategy as proposed in (Foote &
wolffd@0: % Cooper, 2003). First, decompose the file into successive frames of 50 ms 
wolffd@0: % without overlap. 
wolffd@0: help mirframe
wolffd@0: fr = mirframe(a,0.05,1)
wolffd@0: 
wolffd@0: % 3. Compute the spectrum representation (FFT) of the frames.
wolffd@0: sp = mirspectrum(fr)
wolffd@0: clear fr
wolffd@0: % (Remove from the memory any data that will not be used any more.)
wolffd@0: 
wolffd@0: % 4. Compute the similarity matrix that shows the similarity between the 
wolffd@0: % spectrum of different frames.
wolffd@0: help mirsimatrix
wolffd@0: sm = mirsimatrix(sp)
wolffd@0: clear sp
wolffd@0: % Look at the structures shown in the matrix and find the relation with the
wolffd@0: % structure heard when listening to the extract. 
wolffd@0: 
wolffd@0: % 5. Estimate the novelty score related to the similarity matrix. It 
wolffd@0: % consists in a convolution of the diagonal of the matrix with a
wolffd@0: % checker-board Gaussian kernel. Use the novelty function for that purpose. 
wolffd@0: help mirnovelty
wolffd@0: nv = mirnovelty(sm)
wolffd@0: 
wolffd@0: % 6. Detect the peaks in the novelty score.
wolffd@0: help mirpeaks
wolffd@0: p1 = mirpeaks(nv)
wolffd@0: 
wolffd@0: % You can change the threshold value of the peak picker function in order to 
wolffd@0: % get better results. 
wolffd@0: p2 = mirpeaks(nv,'Contrast',0.01)
wolffd@0: 
wolffd@0: clear nv
wolffd@0: 
wolffd@0: % 7. Segment the original audio file using the peaks as position for
wolffd@0: % segmentation. 
wolffd@0: help mirsegment
wolffd@0: s1 = mirsegment(a,p1)
wolffd@0: clear p1
wolffd@0: 
wolffd@0: % 8. Listen to the results.
wolffd@0: mirplay(s1)
wolffd@0: 
wolffd@0: %s2 = mirsegment(a,p2)
wolffd@0: %clear p2
wolffd@0: %mirplay(s2)
wolffd@0: 
wolffd@0: % 9. Compute the similarity matrix of this obtained segmentation, in order
wolffd@0: % to view the relationships between the different segments and their
wolffd@0: % possible clustering into higher-level groups. 
wolffd@0: mirsimatrix(s1,'Similarity')
wolffd@0: clear s1
wolffd@0: %mirsimatrix(s2)
wolffd@0: %clear s2
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: % 10. Change the size of the kernel used in the novelty function, in order
wolffd@0: % to obtain segmentations of different levels of detail, from detailed
wolffd@0: % analysis of the local texture, to very simple segmentation of the whole 
wolffd@0: % piece.
wolffd@0: n100 = mirnovelty(sm,'KernelSize',100)
wolffd@0: n50 = mirnovelty(sm,'KernelSize',50)
wolffd@0: n10 = mirnovelty(sm,'KernelSize',10)
wolffd@0: clear sm
wolffd@0: % As you can see, the smaller the gaussian kernel is, the more peaks can be
wolffd@0: % found in the novelty score. Indeed, if the kernel is small, the cumulative
wolffd@0: % multiplication of its elements with the superposed elements in the
wolffd@0: % similarity matrix may vary more easily, throughout the progressive
wolffd@0: % sliding of the kernel along the diagonal of the similarity matrix, and
wolffd@0: % local change of texture may be more easily detected. On the contrary,
wolffd@0: % when the kernel is large, only large-scale change of texture are
wolffd@0: % detected.
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: p100 = mirpeaks(n100,'NoBegin','NoEnd')
wolffd@0: clear n100
wolffd@0: p50 = mirpeaks(n50,'NoBegin','NoEnd')
wolffd@0: clear n50
wolffd@0: p10 = mirpeaks(n10,'NoBegin','NoEnd')
wolffd@0: clear n10
wolffd@0: s100 = mirsegment(a,p100)
wolffd@0: clear p100
wolffd@0: mirplay(s100)
wolffd@0: clear s100
wolffd@0: s50 = mirsegment(a,p50)
wolffd@0: clear p50
wolffd@0: mirplay(s50)
wolffd@0: clear s50
wolffd@0: s10 = mirsegment(a,p10)
wolffd@0: clear p10
wolffd@0: mirplay(s10)
wolffd@0: clear s10
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: % One more compact way of writing these commands is as follows:
wolffd@0: mirsegment(a,'Novelty')
wolffd@0: mirsegment(a,'Novelty','Contrast',0.01)
wolffd@0: mirsegment(a,'Novelty','KernelSize',100)
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: % Besides, if you want to see the novelty curve with the peaks, just add a
wolffd@0: % second output:
wolffd@0: [s50 p50] = mirsegment(a,'Novelty','KernelSize',50)
wolffd@0: clear s50 p50
wolffd@0: [s10 p10] = mirsegment(a,'Novelty','KernelSize',10)
wolffd@0: clear a s10 p10
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: 
wolffd@0: % 11. Try the whole process with MFCC instead of spectrum analysis. Take the
wolffd@0: % first ten MFCC for instance.
wolffd@0: help mirsegment
wolffd@0: % The segment function can simply be called as follows:
wolffd@0: sc = mirsegment('czardas','Novelty','MFCC','Rank',1:10)
wolffd@0: clear sc
wolffd@0: 
wolffd@0: % Here are some other examples of use:
wolffd@0: [ssp p m b] = mirsegment('valse_triste_happy','Spectrum',...
wolffd@0:                                 'KernelSize',150,'Contrast',.1)
wolffd@0: clear p m b
wolffd@0: mirplay(ssp)
wolffd@0: clear ssp
wolffd@0: 
wolffd@0: display('Strike any key to continue...');
wolffd@0: pause
wolffd@0: close all
wolffd@0: 
wolffd@0: [smfcc2 p m a] = mirsegment('valse_triste_happy','MFCC',2:10,...
wolffd@0:                                 'KernelSize',150,'Contrast',.1)
wolffd@0: clear p m a
wolffd@0: mirplay(smfcc2)