| 摘要 |
FIELD: information technology. ^ SUBSTANCE: invention is meant for audio information input. The system has a microphone, connected to a speech sensor, which generates a signal based on an input non-audio signal, generated based on an effect produced by a speaking person such as facial movement, bone vibration, throat vibration, change of internal impedance of the throat etc. The component of the speech detector generates a speech detection signal, which indicates the probability of what the user is saying, and further calculates the composite signal by multiplying the speech detection signal by the microphone signal. A speech recognition processor recognises speech so as to provide for an output recognition signal, indicating speech in the microphone signal, based on the composite signal. ^ EFFECT: suppression of noise, related to background speech and user friendliness. ^ 13 cl, 11 dwg |
