Speech enhancement method and device thereof, equipment and medium

Abstract

The invention discloses a speech enhancement method and a device thereof, equipment and a medium. The method comprises the following steps: acquiring a target noisy voice signal and performing short-time Fourier transform on the target noisy voice signal to obtain a target frequency domain signal corresponding to the target noisy voice signal; inputting the target feature of the current signal frame of the target frequency domain signal into an encoder in a voice noise suppression model obtained by pre-training to obtain an encoding feature corresponding to the current signal frame of the target frequency domain signal; inputting the coding feature and a decoding feature corresponding to a previous signal frame of a current signal frame of a target frequency domain signal output by a decoder in a voice noise suppression model into the decoder to obtain a decoding feature corresponding to the current signal frame of the target frequency domain signal; and performing signal reconstruction on the decoding features corresponding to each signal frame of the target frequency domain signal to obtain a target enhanced voice signal corresponding to the target noisy voice signal. According to the technical scheme, the speech enhancement effect can be improved, and calculation time and calculation cost are reduced.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of signal processing, and in particular, to a speech enhancement method, device, equipment, and medium. Background technique [0002] The task of speech enhancement is to maximize the perceived quality of the speech signal and suppress the interference of background noise. Speech enhancement technology is generally based on the frequency domain signal of the speech signal or the signal characteristics of the speech signal. In the traditional method, the methods used for speech enhancement mainly include: spectral subtraction, Wiener filtering method, least quadratic method based on statistical features Most of these algorithms deal with limited conditions of noise types and rely on first-order statistical properties. To circumvent the limitations in these algorithms, deep networks have been increasingly used in noise suppression problems. [0003] At present, the methods of deep netwo...

AI Technical Summary

Problems solved by technology

In speech enhancement applications, other deep networks can also be used, for example, convolutional neural network (Convolutional Neural Network, CNN), deep neural network (Deep Neural Networks, DNN) and recurrent neural network (Recurrent Neural Network, RNN), etc. , but CNN and DNN can only process the frequency domain signal corresponding to the speech signal frame by frame, resulting in unsatisfactory speech enhancement effects, and because the speech signal itself has the characteristics of a large amount of data, the RNN and GAN methods are limited by recursion Calculation, unable to perform parallel calculations, resulting in long calculation time and high calculation costs

Images