This paper deals with the application of Higher Order Statistics (HOS) in speech coding. A new model for speech production is implemented and the parameters of the vocal track are obtained from a linear combination of third order cumulants of the speech signal. Noisy signals have been used to test the system including colored Gaussian noise and sinusoidal interfering signals. Pitch and voicing decision are also obtained by HOS analysis of speech.
Keywords: Cumulants, Speech coding