Mixed voice separation method and device, storage medium and electronic equipment

Abstract

The invention discloses a mixed voice separation method and device based on artificial intelligence, a storage medium and electronic equipment. The method comprises the following steps: acquiring a time domain waveform of a mixed voice to be separated; obtaining a 3D tensor of the time domain waveform; using the 3D tensor as input data of a first-layer processing structure in the N-layer processing structures, and inputting into the N-layer processing structures, wherein when the first N / 2-layer processing structures except the first-layer processing structure in the N-layer processing structures are subjected to resampling, the sampling frequency is smaller and smaller, and the sampling data sampled each time is larger and larger; and when the rear N / 2-layer processing structures in the N-layer processing structures are resampled, the sampling frequency is larger and larger, and the sampling data sampled each time is smaller and smaller; obtaining target output data of the last layer of processing structure of the N-layer processing structures; and separating the target output data into multiple paths of voices. According to the invention, the technical problem of low mixed voice separation efficiency is solved.

Description

technical field [0001] The invention relates to the field of computers, in particular to a hybrid speech separation method and device, a storage medium and electronic equipment. Background technique [0002] In the prior art, after the mixed speech is obtained, in the process of separating the mixed speech, the trained neural network model can be used to complete the separation task, and the separated speech can be obtained. However, when the traditional neural network model performs the separation task to separate mixed speech, it needs to use a large number of samples for a large amount of pre-training in order to obtain a separation model that can be used. [0003] That is to say, in the prior art, there is a problem of low separation efficiency in the process of separating mixed voices. [0004] For the above problems, no effective solution has been proposed yet. Contents of the invention [0005] Embodiments of the present invention provide a hybrid speech separatio...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a hybrid speech separation method and device, a storage medium, and electronic equipment to at least solve the technical problem of low hybrid speech separation efficiency

Images