METHOD FOR THE VOICE RECOGNITION OF A SPEAKER USING A PREDICTIVE MODEL, PARTICULARLY FOR ACCESS CONTROL APPLICATIONS

摘要

The invention discloses a method for the voice recognitionof a speaker using a q-order predictive model, comprisinga step for extracting the statistical characteristics including astep of digital acquisition of the speaker's voice sample,corresponding to one or several utterances, a step of convertingthis voice sample into a sequence of vectors of size p,obtained from a series of analysis windows of average size Tand an average spacing I, and a step of determining q+1matrices from this vector sequence. The average size T is ofa duration of less than 10 ms and the average spacing I ofa duration of less than 4.5 ms. The invention is useful in asound lock including an electroacoustical conversion system(HP, 6) and a recorded programme implementation system(5) for the method.