A Zero-Pole Vocal Tract Model Estimation Method Accurately Reproducing Spectral Zeros

DAMIÁN MARELLI; PETER BALAZS

Viena

Congreso; Audio Engineering Society (AES) Convention; 2007

Audio Engineering Society (AES)

Model based speech coding consists in modeling the vocal tract as a linear time-variant system.The all-pole model produced by the computationally efficiency Linear Predictive Coding methodprovides a good representation for the majority of speech sounds. However, nasal and fricativesounds, as well as stop consonants, contain spectral zeros which requires the use of a zero-polemodel. Roughly speaking, a zero-pole model estimation method typically does a non-parametricestimation of the vocal tract impulse response, and tunes the zero-pole model to fit this estimationin a square sense. In this paper we propose an alternative strategy. We tune the zero-pole modelto directly fit the power spectrum of the speech signal in a logarithmic scale, to be consistent withthe way the human ear perceives sounds. In this way, we avoid the error introduced by the vocaltract impulse response estimation and obtain a model that is more accurate at reproducing spectralzeros in a logarithmic scale. A drawback of the proposed method, however, is its computationalcomplexity.