Speech Prosody 2008

Campinas, Brazil
May 6-9, 2008

Unsupervised Prosodic Break Detection in Mandarin Speech

Jui-Ting Huang (1), Mark Hasegawa-Johnson (1), Chilin Shih (2)

(1) Department of Electrical and Computer Engineering; (2) Departments of EALC/Linguistics, University of Illinois at Urbana-Champaign, USA

We propose that, in Mandarin speech, an automatic prosodic break detector can be trained without any prosodically labeled training data. We use only lexical and acoustic cues to create a small labeled training set, then use semi-supervised learning to train a prosodic break detector. A generative mixture model is proposed as the learning algorithm that can learn with both labeled and unlabeled data. The experiments in both English and Mandarin corpus verify our algorithm.

Full Paper

Bibliographic reference.  Huang, Jui-Ting / Hasegawa-Johnson, Mark / Shih, Chilin (2008): "Unsupervised prosodic break detection in Mandarin speech", In SP-2008, 165-168.