Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Total Least Squares Based Subband Modelling for Scalable Speech Representations with Damped Sinusoids

Kris Hermus, Werner Verhelst, Patrick Wambacq, Philippe Lemmerling

Katholieke Universiteit Leuven - ESAT/PSI, Leuven, Belgium

We describe how Total Least Squares (TLS) algorithms can be applied as a powerful and e∆cient modelling tool for wideband speech. A detailed description in both time domain and fre- quency domain illustrates how the modelling functions { damped sinusoids { naturally synthesise non-stationary signals. Straightforward implementations of TLS applied to fullband speech are known to be computationally hard and they can suffer from numerical sensitivity.

In this paper we introduce a subband approach, which leads to a significant reduction of the computational load with an enhanced numerical stability. Moreover, it enables to control the distribution of the TLS components over the spectral range of the input signal such that perceptual criteria can be incorporated in the modelling scheme.

We also address the scalability of our design from smallband speech to high quality audio, and provide evidence for the existence of coupled components in TLS modelled segments.

Full Paper

Bibliographic reference.  Hermus, Kris / Verhelst, Werner / Wambacq, Patrick / Lemmerling, Philippe (2000): "Total least squares based subband modelling for scalable speech representations with damped sinusoids", In ICSLP-2000, vol.3, 1129-1132.