SPEAKER RECOGNITION METHOD

摘要

The invention relates to a method for the automatic recognition of a speaker from a first speech signal that is known to be from the speaker, a second speech signal corresponding to forensic evidence or a test and a plurality of speech signals corresponding to a reference population or selected cohort, comprising the generation of a set of partial models having the same duration as that of the signals corresponding to the selected cohort, such that the above-mentioned signals are used systematically in order to determine if the speaker of the first known signal is the same as the speaker of the second test signal.