METHOD FOR GENERATING A REDUCED AUDIO IMPRINT FROM A SOUND SIGNAL AND METHOD FOR IDENTIFYING A SOUND SIGNAL USING SUCH A REDUCED AUDIO IMPRINT

摘要

The present invention relates to a process for the generation of an audio reduced from a sound signal x (t), comprising the following steps: a) cut the sound signal into a succession of frames; b) to generate a descriptor tonal for each frame, so as to associate with each frame a vector of real numbers and dimension c; c) control the vector; d) divided by a statistical analysis the vectors in m classes denoted cm, m being an integer varying from 1 to m, e) assigned to each vector is an integer m, m being the index of the class cm which is part of the vector so that the sound signal is represented by an ordered sequence of integers constituting the impression audio reduced. The present invention also relates to a system and method of identification of a sound signal, not by comparing the imprint of the audio reduced and those contained in a reference database.