摘要 |
The invention relates to a device and a method at a speech verification system which contributes to an improved identification of the received speech by analysis of the internal relations of time between the formant movements. The device according to invention analyses the point of time when the different formants reach their maxima respective minima regarding said resonance frequency values, at which said values are utilized to establish a speaker-specific frequency profile which is stored in a speech verification database. When a speaker's identity shall be verfied, for instance, to get access to a certain telecommunication service, then intended units in the device compare the received speech with in said database stored frequency profile. <IMAGE> |