Paper: Speech Summarization Without Lexical Features for Mandarin Broadcast News

ACL ID N07-2054
Title Speech Summarization Without Lexical Features for Mandarin Broadcast News
Venue Human Language Technologies
Session Short Paper
Year 2007
Authors

We present the first known empirical study on speech summarization without lexi- cal features for Mandarin broadcast news. We evaluate acoustic, lexical and struc- tural features as predictors of summary sentences. We find that the summarizer yields good performance at the average F- measure of 0.5646 even by using the com- bination of acoustic and structural features alone, which are independent of lexical features. In addition, we show that struc- tural features are superior to lexical fea- tures and our summarizer performs sur- prisingly well at the average F-measure of 0.3914 by using only acoustic features. These findings enable us to summarize speech without placing a stringent demand on speech recognition accuracy.