Why Corpora for Continuous Speech?

The answer is very simple: only when we study continuous speech can we find fluent speech prosody. We have noted that our community have focused primarily on studying segments, tones, and individual phrases or sentences, and tended to treat continuous speech as strings of discrete or UNRELATED sentences or phrases.

F0 contour over multiple phrases. With COSPRO, we would like to bring your attention to the existence of multiple phrase speech paragraphs in narratives and spoken discourses, and in particular, the prosodic relationships between and among the phrases within and across speech paragraphs. This perspective has led to how fluent speech should be decomposed and oranized systematically. In other words, how multiple phrase prosody is formed on top of and in addition to intonation types by phrase or sentnece.

Sinica COSPRO is by far the only and largest corpora of Mandarin Chinese continuous speech designed to bring out features of fluent speech prosody. It is designed, collected and annotated by Dr. Chiu-yu Tseng and her research group at the Phonetics Lab, Institute of Linguistics, Academia Sinica, Taipei, Taiwan.

資料來源:http://reg.myet.com/registration/corpus/en/Main.asp