From: Voice-to-Phoneme Conversion Algorithms for Voice-Tag Applications in Embedded Platforms
 | Word accuracy (%) with CD HMMs | |||
---|---|---|---|---|
Sequentially combine hypotheses from | Single sample | Two samples | Three samples | Baseline |
Monogram histogram | 87.3 | 93.9 | 95.6 | 92.69 |
Bigram Histogram | 87.3 | 94.0 | 95.2 | 92.69 |