Abstract
The mechanism of speech production imposes a number of well-defined constraints on the resulting acoustic wave. The nature of these constraints is associated with the anatomy of the human vocal tract and vocal cords. This mechanism can be characterized as a time-varying linear dynamic system excited alternatively by a noise or pulse source which contains only frequencies much higher than the variational rate. Thus, speech can be described by a small number of parameters which change at the slower rate. Such parameters in effect describe the constants of the dynamic system and its excitation. Representations of this kind demonstrate the philosophy underlying the signal theory approach to the design and analysis of transmission circuits.

This publication has 26 references indexed in Scilit: