In this study, I applied wavelet transform to speech analysis and showed its effectiveness to speech analysis. Wavelet transforms can detect frequency and positional information at the sane time thou Fourier Transforms cannot. Wavelet transforms are useful for detecting linguistic sound features which have their rapid spectral change in time. Wavelet transforms are applied to consonantal feature analysis of Chinese Mandarin aspirated/unaspirated labials [pa,p{sup}(ha)]. I will try to show that wavelet transforms are more useful than Fourier-based spectrograph for speech analysis.
展开▼