Abstract | ||
---|---|---|
We present a Mandarin keyword search system that uses a large vocabulary recognizer to generate consensus networks at various resolutions: word, character, syllable and phone. In order to achieve fast and accurate search, we propose the use of an efficient approximate-match dynamic programming algorithm that finds the best alignment between the target query and the consensus network. Experiments with Mandarin conversational telephone speech show that the approximate-match search improves detection accuracy by more than 10% for rare words that are not present in the recognizer's dictionary (OOV terms). We also found OOV terms to benefit most from system combination, where we observe a roughly 10% improvement relative to the best single system. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1109/ICASSP.2012.6289087 | ICASSP |
Keywords | Field | DocType |
dynamic programming,natural language processing,speech processing,Mandarin conversational telephone speech,OOV terms,approximate-match dynamic programming algorithm,consensus network,spoken term detection,target query,unseen words detection,Mandarin,OOV,Spoken term detection | Speech processing,Dynamic programming,System combination,Computer science,Decision support system,Speech recognition,Phone,Syllable,Artificial intelligence,Natural language processing,Vocabulary,Mandarin Chinese | Conference |
ISSN | Citations | PageRank |
1520-6149 | 4 | 0.89 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ivan Bulyko | 1 | 249 | 22.40 |
Owen Kimball | 2 | 83 | 17.82 |
Manhung Siu | 3 | 464 | 61.40 |
Jose Herrero | 4 | 4 | 0.89 |
Dan Blum | 5 | 4 | 0.89 |