A Robust HMM-Based Endpoint Detector for Telecommunication Applications
- Alex Acero
Proc. of Eurospeech |
Published by International Speech Communication Association
A new real time HMM-Based endpoint detector is proposed in this paper. Endpoint detection has been shown to be critical in automatic speech recognition systems. The system uses static (energy) and dynamic (delta energy) features of the signal on a frame by frame basis. The endpoint detector is trainable for the working conditions (i.e. telephone lines) and is able to track changes in background noise conditions. Our experiments indicate that high accuracy, low false rejection and low false alarm rates can be obtained with this new endpoint detector.
© 2007 ISCA. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the ISCA and/or the author.