Automatic Phonetic Segmentation using Boundary Models

Jiahong Yuan, Neville Ryant, Mark Liberman, Andreas Stolcke, Vikramjit Mitra, and Wen Wang

Abstract

This study attempts to improve automatic phonetic segmentation within the HMM framework. Experiments were conducted to investigate the use of phone boundary models, the use of precise phonetic segmentation for training HMMs, and the difference between context-dependent and contextindependent phone models in terms of forced alignment performance. Results show that the combination of special one-state phone boundary models and monophone HMMs can significantly improve forced alignment accuracy. HMM-based forced alignment systems can also benefit from using precise phonetic segmentation for training HMMs. Context-dependent phone models are not better than context-independent models when combined with phone boundary models. The proposed system achieves 93.92% agreement (of phone boundaries) within 20 ms compared to manual segmentation on the TIMIT corpus. This is the best reported result on TIMIT to our knowledge.

Details

Publication typeInproceedings
Published inProceedings Interspeech
PublisherInternational Speech Communication Association
> Publications > Automatic Phonetic Segmentation using Boundary Models