Table 8 Recognition results on the evaluation set using bigram models in the first pass and context-dependent models in the second pass
FLMs | Optimal model order | WER [%] | Relative improvement[%] |
|---|---|---|---|
P | 5 | 26.51 | 9.26 |
M | 4 | 25.84 | 11.56 |
W | 5 | 26.07 | 10.78 |
P+M | 5 | 25.92 | 11.27 |
P+W | 5 | 26.38 | 9.71 |
M+W | 4 | 25.92 | 11.30 |
All | 5 | 26.01 | 10.99 |