INTERSPEECH
2010
13 years 6 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...