Abstract | ||
---|---|---|
This paper presents a novel bilingual model modification approach to improve normative speech recognition accuracy when the variations of accented pronunciations occur. Each state of baseline normative acoustic model is modified with several candidate states from the auxiliary acoustic model, which is trained on speakers' mother language. State mapping criterion and n-best candidates are investigated, and different numbers of Gaussian mixture's of the auxiliary acoustic model are compared based on a grammar-constrained speech recognition system. Using this bilingual model modification approach, compared to the normative acoustic model which has already been well trained by adaptation technique MAP, the Phrase Error Rate further achieves a 5.83% relative reduction, while only a small relative increase on Real Time Factor occurs. |
Year | DOI | Venue |
---|---|---|
2009 | 10.1007/978-3-642-01216-7_32 | SIXTH INTERNATIONAL SYMPOSIUM ON NEURAL NETWORKS (ISNN 2009) |
DocType | Volume | ISSN |
Conference | 56 | 1867-5662 |
Citations | PageRank | References |
0 | 0.34 | 3 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Qingqing Zhang | 1 | 102 | 14.76 |
Jielin Pan | 2 | 44 | 18.04 |
Shui-duen Chan | 3 | 5 | 1.57 |
Yonghong Yan | 4 | 656 | 114.13 |