Vocal Tract Estimation in High Pitched Voices

Vocal tract estimation is highly useful for many speech signal processing applications. Earlier methods to do these involved many spectral methods such as Linear predictive Coding, Cepstral Smoothing of amplitude spectrum and other interpolation schemes. However such methods are limited to the cases of lower pitches wherein the amplitude spectrum has sufficient information. At higher pitches it is necessary to search for different methods to carry out this vocal tract estimation. In this project we implement various methods already available and we try to develop our own method to improve the performance.