| 摘要 |
An acoustic input (90) is recognized from inferred articulatory movements output by a learned relationship between training acoustic waveforms and articulatory movements. The inferred movements are compared (98) with template patterns (46) prepared from training movements when the relationship was learned to regenerate an acoustic recognition (99). In a preferred embodiment, the acoustic-articulatory relationships are learned by a neural network. Subsequent input acoustic patterns then generate the inferred articulatory movements for use with the templates. Articulatory movement data may be supplemented with characteristic acoustic information, e.g. relative power and high frequency data (22), to improve template recognition. |
