A prosody-only decision-tree model for disfluency detection (1997)

Content Provider	CiteSeerX
Author	Shriberg, Elizabeth Bates, Rebecca Stolcke, Andreas
Description	PROC. EUROSPEECH
Abstract	Speech disfluencies (filled pauses, repetitions, repairs, and false starts) are pervasive in spontaneous speech. The ability to detect and correct disfluencies automatically is important for effective natural language understanding, as well as to improve speech models in general. Previous approaches to disfluency detection have relied heavily on lexical information, which makes them less applicable when word recognition is unreliable. We have developed a disfluency detection method using decision tree classifiers that use only local and automatically extracted prosodic features. Because the model doesn’t rely on lexical information, it is widely applicable even when word recognition is unreliable. The model performed significantly better than chance at detecting four disfluency types. It also outperformed a language model in the detection of false starts, given the correct transcription. Combining the prosody model with a specialized language model improved accuracy over either model alone for the detection of false starts. Results suggest that a prosody-only model can aid the automatic detection of disfluencies in spontaneous speech. 1.
File Format	PDF
Publisher Date	1997-01-01
Access Restriction	Open
Subject Keyword	Automatic Detection Effective Natural Language Understanding Speech Model Word Recognition Specialized Language Model Disfluency Type Lexical Information Disfluency Detection Prosody-only Decision-tree Model Language Model Disfluency Detection Method Prosodic Feature Prosody-only Model Correct Transcription Prosody Model Decision Tree Classifier Spontaneous Speech Previous Approach Speech Disfluency Model Doesn False Start
Content Type	Text