Sound denoising method and device

Abstract

The embodiment of the invention provides a sound denoising method and device, and relates to the technical field of audio processing. The method comprises the steps: obtaining first sound data of a current sampling point; obtaining second sound data, wherein the second sound data is the data collected at second sampling moments, the second sampling moments are a preset number of sampling moments before the first sampling moment and are the closest to the first sampling moment, and the first sampling moment is the sampling moment corresponding to the first sound data; carrying out the sliding smooth processing of the first sound data according to the second sound data; and determining the processed sampling data as the data after the denoising of the first sound data. According to the technical scheme of the embodiment, the method can remove the noise of an abrupt change type in a sound signal, and improves the sound quality.

Description

technical field [0001] The present application relates to the technical field of audio processing, in particular to a sound noise reduction method and device. Background technique [0002] When an audio collection device collects a sound signal, it is easily affected by surrounding environmental factors, so that the collected sound signal contains noise, which affects the quality of the sound. There are many sources of these noises, including equipment interference and electromagnetic interference, for example, noise generated by motor rotation and noise generated by electromagnetic signal interference. The noise produced can also be divided into many types, which can include sound energy peaks that are not large, energy changes slowly, and last for a long time. Short bursts of noise. figure 1 A time-domain diagram of noisy speech data, which contains both abrupt noise and stationary noise. [0003] In the prior art, spectral subtraction is often used to reduce the noise ...

AI Technical Summary

Problems solved by technology

[0005] Among them, the stationary noise has relatively obvious spectral characteristics, while the abrupt noise does not have obvious spectral characteristics, so when the collected noise samples contain both stationary noise and abrupt noise, it is difficult to obtain from the noise samples Spectrum of abrupt noise, for example, figure 2 It is a frequency-domain diagram of speech data that mainly carries sudden noise. It can be seen from the figure that the sudden noise distribution is very dense, there is no obvious frequency characteristic, and the frequency spectrum of sudden noise cannot be estimated.

[0006] Based on the above, the spectral characteristics of stationary noise are more prominent. When the above method is used to denoise the sound signal, the stationary noise can be removed better, but it is difficult to remove the sudden noise, which will easily lead to low sound quality after noise reduction.

Images