The long term research objective is to develop means of separating the speech waveform into voice source and vocal tract components. From the voice source, the goal is to determine the operation of the vocal folds; from the vocal tract component, the goal is to derive articulation. This objective addresses the fundamental problem of voice source-vocal tract transfer function ambiguity which impedes the determination of the underlying physiological source of many speech disorders. The proposed research will concentrate on the ambiguity problem from the standpoint of the voice source. Inverse filtering will be evaluated as a technique for estimating the vocal tract transfer function from the speech wave and for removing the effects of this transfer function to reveal the underlyin glottal airflow wave--the voice source the inverse filter glottal wave estimate will be compared against a computer model reconstruction of the glottal wave based on measurements of human speech. These measurements include stroboscopic measurement of glottal motion waveforms, electroglottographic waveforms, acoustic pulse reflection reconstructions of the vocal tract area function, and the acoustic speech wave.
|Forrest, K; Weismer, G; Milenkovic, P et al. (1988) Statistical analysis of word-initial voiceless obstruents: preliminary data. J Acoust Soc Am 84:115-23|
|Milenkovic, P (1987) Least mean square measures of voice perturbation. J Speech Hear Res 30:529-38|