| 摘要 |
A speech recognition apparatus for use in an environment containing ambient acoustic noise includes a speech recogniser (20) for receiving and processing data representative of a speech utterance to be recognised to output data representative of or dependent on the lexical content of said utterance. The apparatus includes level adjusting means (16) for adjusting the level of the speech utterance. The apparatus also includes means for measuring said ambient acoustic noise level before or during said utterance, processing means (22) for using said measured acoustic noise level to predict the likely level of the speech utterance, and for adjusting the gain of the signal in accordance with said prediction of the likely level of the speech utterance. |
