Quick audio-frequency separating method based on tonic frequency

The present invention relates to a quick audio segmentation method based on pitch frequency, pitch frequency detected by first speech from the audio, and then the energy of the non-speech portion into mute and un-mute, speaker finally detected change between speech segments, by the speaker people change point to get the final segmentation result. Advantageous effects of the present invention are: the use of the pitch frequency to segment the prosodic feature mute, voice and non-voice, non-voice eliminate the negative impact on the division of the speaker; the pitch frequency more than cepstral feature saves computing time.