Voice enhancement network model and single-channel speech enhancement method and system

Abstract

The invention provides a single-channel speech enhancement method. The method is realized through a speech enhancement network model. The speech enhancement network model comprises an analysis layer,an encoder, a time convolution module, a decoder and a synthesis layer. According to the single-channel speech enhancement method provided by the invention, an analysis layer of quasi-short-time windowed Fourier transform based on convolution layer design and a synthesis layer of quasi-inverse short-time windowed Fourier transform are added, so that the characteristics of speech are better mined in a transform domain. Besides, the encoder and the decoder are constructed by adopting the gating convolution layer so as to expand the receptive field, transmission of information in the hierarchicalstructure can be better controlled, and the time convolution module is added between the encoder and the decoder so as to better learn the long-term memory characteristic of the speech, so that the speech enhancement effect can be enhanced. Meanwhile, the invention provides a single-channel speech enhancement system and a speech enhancement network model.

Description

technical field [0001] The invention relates to the technical field of speech signal processing, in particular to a single-channel speech enhancement method, a single-channel speech enhancement system and a speech enhancement network model. Background technique [0002] Speech enhancement refers to the use of audio signal processing technology and various algorithms to improve the intelligibility or overall perceptual quality of distorted speech signals, thereby further improving application effects in speech recognition, voice calls, hearing aids, and voiceprint recognition. Traditional single-channel speech enhancement algorithms include spectral subtraction, algorithms based on statistical models, and subspace algorithms. However, such algorithms usually need to make certain assumptions about the respective characteristics of the speech signal and noise and whether they are related to each other, which limits their enhancement performance. Most of the speech enhancement ...

AI Technical Summary

Problems solved by technology

However, such algorithms usually need to make certain assumptions about the characteristics of the speech signal and noise and whether they are related to each other, which limits their enhancement performance.

Most of the speech enhancement algorithms based on deep learning currently use frequency-domain features, such as short-time Fourier transform magnitude spectrum or logarithmic power spectrum, and the phase of the enhanced speech is replaced by the phase of the noisy speech, making the enhancement There is a certain mismatch between phase and amplitude in the speech, resulting in the "inconsistent spectrum" problem

Images