Voice and non-voice detection method based on fractional order Fourier transformation

The invention discloses a voice and non-voice detection method based on fractional order Fourier transformation, and relates to the field of voice signal processing. According to the technical scheme, the method includes the steps that at first, framing is carried out on audio signals, zero equalization and windowing processing are carried out, fractional order Fourier transformation of different orders is conducted on each frame of signal, a fractional order frequency spectrum of the signals is obtained, then fractional order frequency spectrum entropies are calculated, the maximum fractional order frequency spectrum entropy of each order is selected, the maximum fractional order frequency spectrum entropies are compared with a threshold value, and the judgment result can be obtained. The method is unified in processing, simple in procedure and capable of effectively improving voice and non-voice detection performance; trashy signals such as mute, noise and music in the audio signals can be effectively filtered.