A primary difference between spontaneous speech and read speech concerns the presence of disfluencies, which are much more prevalent in spontaneous speech. Disfluencies often take the form of hesitation pauses (both filled and unfilled) and false starts. With false starts, the speaker interrupts the normal flow of speech to restart an utterance. With pauses, extraneous silences or "uh"-type sounds are inserted. The acoustic aspects of such restarts and pauses in a widely-used speech database were examined here, from the point of view of identifying them acoustically. Automatically locating such disfluencies could improve the performance of an automatic speech recognizer, by allowing the elimination from consideration of some hypotheses based on spectral analysis. Without such analysis, disfluent speech risks being misinterpreted by a recognizer.

Keywords: Restarts, pauses, recognition, intonation