Blind source separation method applied to voice noise reduction

The invention discloses a blind source separation method applied to voice noise reduction in the technical field of blind source separation, and the method comprises the steps: carrying out the first blind source separation operation of an unknown strong interference signal in a source mixed signal, and obtaining the parameter characteristics of the unknown strong interference signal, and comprises the steps: obtaining the multi-channel audio data of an environment where a target sound source is located; separating the multi-channel audio data based on a trained separation model to obtain single-channel audio data; and taking the single-channel audio data as audio data of a target sound source. According to the method, the problem of voice overlapping when a plurality of speakers speak in the same time period is solved, voices and contents when different speakers speak in the same time period can be accurately segmented, a two-dimensional feature can be obtained by convolution of multi-channel audio data and