Estimation of vocal tract shape during stop closures

The display of intensity, pitch, and vocal tract shape is considered to be helpful in speech training of the hearing impaired. Vocal tract shape estimation, based on LPC and other analysis techniques, works satisfactorily for vowels but generally fails during stop consonants due to very low signal energy. Our estimation of vocal tract shape is based on reflection coefficients obtained from LPC analysis of speech. In order to study the consistency and dynamics of the vocal tract shape estimation with amplitude variation of vowels, and to study the transitions at vowel-consonant boundaries, we have used “areagram”, a spectrogram-like two-dimensional display of estimated vocal tract cross-sectional area as a function of time and position along the tract length. Display of areagram for vowel-consonant-vowel (V-C-V) syllables is satisfactory for vowel part, but during stop closure, area values become random. Based on estimated vocal tract cross sectional area before and after stop closure, conic surfaces are obtained and used for estimation of vocal tract shape during stop closure by a two-dimensional interpolation. Keywords Vocal tract shape estimation, speech training aids for hearing impaired children, place of stop closure. 1