A segmental HMM is a HMM whose states are associated with sequences of acoustic feature vectors (or segments), rather than individual vectors. By treating segments as homogeneous units it is possible, for example, to develop better models of speech dynamics. This paper begins by describing a type of segmental HMM in which the relationship between the state and acoustic level descriptions of a speech signal is regulated by an intermediate, articulatory layer, and discusses its potential benefits for speaker recognition. As a first step towards applying this type of model to speaker recognition, text-dependent speaker verification results obtained on YOHO using a simpler segmental HMM are presented, which show a 44% reduction in false acceptances using the segmental model compared with a conventional HMM. Experiments in text-independent speaker verification on Switchboard are then described.
Cite as: Liu, Y., Russell, M., Carey, M. (2004) Speaker recognition using a trajectory-based segmental HMM. Proc. The Speaker and Language Recognition Workshop (Odyssey 2004), 45-50
@inproceedings{liu04_odyssey,
author={Ying Liu and Martin Russell and Michael Carey},
title={{Speaker recognition using a trajectory-based segmental HMM}},
year=2004,
booktitle={Proc. The Speaker and Language Recognition Workshop (Odyssey 2004)},
pages={45--50}
}