Cross-task portability of a broadcast news speech recognition system
Autor: | Mauro Cettolo, M. Federico, N. Bertoldi, Fabio Brugnara, Diego Giuliani |
---|---|
Rok vydání: | 2002 |
Předmět: |
Linguistics and Language
Voice activity detection business.industry Computer science Communication Speech recognition Acoustic model Speech corpus computer.software_genre VoxForge Speech processing Porting Language and Linguistics Computer Science Applications Task (project management) Modeling and Simulation Computer Vision and Pattern Recognition Artificial intelligence Language model business computer Software Natural language processing |
Zdroj: | Speech Communication. 38:335-347 |
ISSN: | 0167-6393 |
Popis: | This paper reports on experiments of porting the ITC-irst Italian broadcast news recognition system to two spontaneous dialogue domains. Porting was investigated by applying state-of-the-art adaptation methods on acoustic and language models, and by evaluating the trade-off between performance and required amount of task specific annotated data. The use of different levels of supervision for acoustic model adaptation was also studied. By employing 2 h of manually annotated speech, word error rates of 26.0% and 28.4% were achieved by the adapted systems. These results are to be compared with the performance of two domain specific baseline systems, 22.6% and 21.2%, respectively, which were developed on much more training data. Finally, a robust method is presented that allows to tune the insertion of spontaneous speech phenomena by the speech decoder. |
Databáze: | OpenAIRE |
Externí odkaz: |