9th Annual Conference of the International Speech Communication Association

Brisbane, Australia
September 22-26, 2008

On the Properties of a Time-Varying Quasi-Harmonic Model of Speech

Yannis Pantazis (1), Olivier Rosec (2), Yannis Stylianou (1)

(1) University of Crete, Greece; (2) Orange Labs, France

In this paper we present the properties of a parametric speech model based on a deterministic plus noise representation of speech initially suggested by Laroche et al. [1]. Aiming at a high resolution analysis of speech signals for voice quality control (transformation) and assessment, we focus on the deterministic representation and we reveal the properties of the model showing that such a representation is equivalent to a time-varying quasiharmonic representation of speech. Results show that the model is appropriate in estimating accurately linear amplitude modulations and modeling the inharmonicity of speech.


  1. J. Laroche Y. Stylianou and E. Moulines. HNM: A Simple, Effecient Harmonic plus Noise Model for Speech. In Workshop on Appl. of Signal Proc. to Audio and Acoustics (WASPAA), pages 169-172, New Paltz, NY, USA, Oct 1993.

Full Paper

Bibliographic reference.  Pantazis, Yannis / Rosec, Olivier / Stylianou, Yannis (2008): "On the properties of a time-varying quasi-harmonic model of speech", In INTERSPEECH-2008, 1044-1047.