| P00-1073 | distribution-based pruning of | n-gram backoff | language models . Instead of |
| P99-1002 | cost The speech recognizer uses | n-gram backoff | language models estimated on |
| P00-1073 | proposed a novel approach for | n-gram backoff | models pruning : keep n-grams |
| P04-1008 | transducer Gˆ represents an | n-gram backoff | model for the joint probability |
| P10-1046 | by just under 1 % . The Google | n-gram backoff | model is almost as good as backing |
| P00-1073 | method results in a more general | n-gram backoff | model , which resists to domain |
| W10-3602 | occur only once are pruned , and | n-gram backoff | weights are re-normalized after |
| D08-1087 | in this work can be encoded as | n-gram backoff | models , they are applied directly |
| N12-1014 | Modeling a corpus with n-gram counts | n-gram backoff | language models have been used |
| P00-1073 | method results in a more general | n-gram backoff | model , in spite of the domain |
| W05-1104 | StandardNgramModel class can load standard | n-gram backoff | models for scoring , as shown |
| P12-1063 | ngram models ; first a classical | n-gram backoff | model ( Chen and Goodman , 1999 |
