90 likes | 193 Views
Sentence Unit Detection in Conversational Dialog Speech. Add graphic of sound wave file Text transcription of interspersed speaker words without . , ?. Elizabeth Lingg Tejaswi Tennetti Anand Madhavan. Sentence Units. Questions Back Channels Statements. Data Used.
E N D
Sentence Unit Detection in Conversational Dialog Speech Add graphic of sound wave file Text transcription of interspersed speaker words without . , ? Elizabeth Lingg TejaswiTennetti Anand Madhavan Sentence Units Questions Back Channels Statements
Data Used LDC2009T01: Annotated metadata Fisher data Switchboard corpus POS tags Disfluencies marked
Prediction results Final results of predictions with the best features chosen
Effect of POS tags Many graphs showing 1-pre-gram, 2-pre-gram, 1-post-gram, 2-post-gram and all of them together Vary with cross validation bins used? Vary with many classifiers Above on a per-sentence-unit type basis
Effect of special words for backchannel identification Club words like ‘mhm’, ‘oh yeah’ etc into a separate class and see if it helps in predicting backchannel better Effects on other sentence units
Miscellaneous Previous sentence class prediction (faked as well as true) Length of sentence so far or number of words so far (that have not been classified yet)
Prosodic features F0 F0 normalized Pause duration for speaker Energy Length of word Pause length before word Word pitch range Energy Energy normalized
Prosodic features n-gram prosodic features
References Enriching Speech Recognition With Automatic Detection of Sentence Boundaries and Disfluencies, Yang Liu, Elizabeth Shriberg, Andreas Stolcke, Dustin Hillard, Mari Ostendorf and Mary Harper ...