8th European Conference on Speech Communication and Technology

Geneva, Switzerland
September 1-4, 2003


Towards Missing Data Recognition with Cepstral Features

Christophe Cerisara

LORIA, France

We study in this work the Missing Data Recognition (MDR) framework applied to a large vocabulary continuous speech recognition (LVCSR) task with cepstral models when the speech signal is corrupted by musical noise. We do not propose a full system that solves this difficult problem, but we rather present some of the issues involved and study some possible solutions to them. We focus in this work on the issues concerning the application of masks to cepstral models. We further identify possible errors and study how some of them affect the performances of the system.

Full Paper

Bibliographic reference.  Cerisara, Christophe (2003): "Towards missing data recognition with cepstral features", In EUROSPEECH-2003, 3057-3060.