Analysis and Synthesis of Speaker Based Vocal Tract Shape Estimation for Vowels at different Conditions

Abstract

for analysis of speech production and synthesis with the use of synthesized speech from acoustic parameters as the integral tool for articulatory synthesis system at the Haskin s Laboratories that synthesizes speech through controlled speech articulators (ex: Jaw, Tongue, Lips) instead of acoustic variable, where variability is articulations results in variability of vocal tract shape. This regulated vocal tract works as a filter with its own transfer function, many researchers have focused majorly on the various areas including the controlled articulators and its dimensionality and modeling of tongue and lips event though it is an soft tissue. newlineHere, we aim to analyze an approximate phonatory model by improving the existing approach for variety of acoustic parameters namely fundamental frequency (pitch), formant frequency and zero crossing rate individually with aerodynamic simulation which examines the protruding sound wave along with the non-uniform tubular structure of vocal tract, generating patterns of moment of simulated vocal tract articulation and specifying the temporal relation among dynamically defined gestures that leads to a time varying vocal tract filter function and an acoustic wave output. newlineWith the generated acoustic waveform with its respective rejection co-efficients analysis, estimation of vocal tract shape and also the synthesis of input using linear predictive coding is done for the three various conditions namely Indian English vowels at normal recordings, same vowels recorded by consuming ice cold water and with time lapse of five minutes, as an interesting parameter of adjacent formant frequency ratio f2/f1. f3/f2 and f4/f3 are analyzed which will serve as the major necessary tool for the study of variability of vocal tract shape along with their bandwidth to enhance the application of a vocal tract signature and the study of cryotrapic effects to locate the irregularity in vocal tract. Our approach can be used for the better identification of speaker in spite of the mimicry of the speaker as it is speaker independent recognition. newlinexii newlineThe invention of formant spread individuality phonetive distinctiveness are the essential parameter that can be noticed on comparison with inter and intra speaker recognition which is used for forensic applications. Nevertheless, both within and between speakers the variability of vocal tract shape still not well established therefore an attempt is made to synthesize the vocal tract shape in combination with acoustic parameter along with its PAR-COR coefficients made. newline