Reply by May 22, 20072007-05-22
Hi everybody;

I want to tell you what i did and what i need. I made a data base including phonemes like '' -th,-sh, ai,-ea .." i recorded this phonemes in a wav file. I read a text and recorded it in a wav file. I have a system that you can process the speech with fft,dct and wavelet transform than you have two options for matching or comparing two speech datas. This options are DTW and Cross correlation. I want to find out selected phoneme how many times used in this speech. But when i want to use dtw and cross correlation i have a problem because it cant make efficient compare. So I want to determine the speech\nonspeech parts of the speech. and try to compare this speech part to phonemes so find out how many times this phoneme used in this speech. I find a way which is short time average zero crossing rate but only a short expression I couldt find its algorithm(pseudo code). Would someone propose me an algorithm(pseudo code) or a source code(c,c++ or c#) which I can use. thanks in advance.

best regards