| 摘要 |
<p>The invention discloses a speech recognition process and device by means of which, in the reference signal are predefined a plurality of intervals corresponding to each formant, and comprising a first step whereby is determined the maximum value of the signal in a plurality of frequency bands wherein is divided the first interval, and a second step to calculate a parameter indicating the dissimilarity between the power of the speech signal and the reference signal. The acoustic characteristics of the speech signal and of the reference signal are compared without having to compare the two full spectra between each other.</p> |
