A Pitch-Based Rapid Speech Segmentation for Speaker Indexing

Segmentation of continuous audio is important for speaker indexing. The reliability of models in speaker indexing depends much on segmentation. Commonly used method is based on the Bayesian Information Criteria (BIC), but it is not so capable when dealing with short utterances. In this paper, we present a pitch-based speech segmentation method, which can detect frequently speaker changes accurately and rapidly. First, in our algorithm, utterance segments are detected by pitch. Then distances of pitch are computed, and compared with a self-adaptable threshold. Speaker changes are finally decided among utterance segments. We applied our method and three comparative methods on the HUB4-NE broadcast data. Speaker indexing experiments had been taken following each algorithm. We also suggested two indicators as complements of false alarm and missing rate in the evaluation of segmentation. The experiment results show that our algorithm works faster and better, with most of short time speake...

• Type: Conference
• Year: 2005
• Where: ISM
• Authors: Min Yang, Yingchun Yang, Zhaohui Wu
• Keywords: ISM 2005, Speaker Changes, Speaker Indexing, Utterance Segments