118
click to vote
INTERSPEECH
2010
14 years 10 months ago
2010
Continuous speech input for ASR processing is usually presegmented into speech stretches by pauses. In this paper, we propose that smaller, prosodically defined units can be ident...