摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide a system and a method using, in particular, cepstrum noise subtraction with regard to a voice-recognition system and method. <P>SOLUTION: The voice-recognition system uses a first scalar coefficient and a second scalar coefficient and determines the state for restricting the process for a cepstrum feature vector in order to avoid excessive emphasis or subtraction at the cepstrum feature vector so that the process of the cepstrum feature vector is appropriately conducted to improve anti-noise performance of voice recognition. Also the voice-recognition system and method can be applied to any environment, is not complex, and can be incorporated easily into other system to provide a user with highly reliable and stable voice-recognition results. <P>COPYRIGHT: (C)2009,JPO&INPIT</p> |