Loading...
Please wait, while we are loading the content...
Similar Documents
Advances in Mandarin broadcast speech recognition (2007)
| Content Provider | CiteSeerX |
|---|---|
| Author | Hwang, Mei-Yuh Wang, Wen Lei, Xin Zheng, Jing Cetin, Ozgur Peng, Gang |
| Description | in Proc. Int. Conf. on Speech Communication and Technology We describe our continuing efforts to improve the UW-SRI-ICSI Mandarin broadcast speech recognizer. This includes increasing acoustic and text training data, adding discriminative features, incorporating frame-level discriminative training criterion, multiplepass acoustic model (AM) cross adaptation, language model (LM) genre adaptation and system combination. The net effect without LM adaptation was a 24%-64 % relative reduction in character error rates (CERs) on a variety of test sets. In addition, LM adaptation gave us another 6 % of relative CER reduction on broadcast conversations. Index Terms — Mandarin, character error rates, discriminative features, discriminative training, LM adaptation. 1. |
| File Format | |
| Language | English |
| Publisher Date | 2007-01-01 |
| Access Restriction | Open |
| Subject Keyword | Discriminative Training Uw-sri-icsi Mandarin Broadcast Speech Recognizer Relative Reduction Cross Adaptation Language Model Net Effect Frame-level Discriminative Training Criterion Mandarin Broadcast Speech Recognition Test Set Genre Adaptation Relative Cer Reduction Discriminative Feature Multiplepass Acoustic Model Broadcast Conversation Text Training Data Lm Adaptation Character Error Rate Index Term Mandarin System Combination |
| Content Type | Text |
| Resource Type | Article |
