Fourth European Conference on Speech Communication and Technology

Madrid, Spain
September 18-21, 1995

Speaker Recognition with Temporal Transition Models

Haizhou Li, Jean-Paul Haton, Jian Su, Yifan Gong

CRIN-CNRS/INRIA-Lorraine, Campus Scientifique, Vandoeuvre-les-Nancy, France

In this paper, a temporal transition model (TTM) of speech is proposed for speaker recognition and verification. The TTM is introduced to encode the short time dynamics of speech. The issues on the model building, the distance measures and the implementation are addressed. A set of experiments were conducted based on TTM, which gave a 98.9% recognition rate and 99.5% verification rate on a database of 72 French speakers. The fact is confirmed that temporal dynamics of utterance encodes well speaker specificity.

Full Paper

Bibliographic reference.  Li, Haizhou / Haton, Jean-Paul / Su, Jian / Gong, Yifan (1995): "Speaker recognition with temporal transition models", In EUROSPEECH-1995, 617-620.