Audio noise reduction method and training method of audio noise reduction model

Abstract

The embodiment of the invention discloses an audio noise reduction method and a training method of an audio noise reduction model. According to the scheme, the method comprises the steps: acquiring the audio signal to be denoised, processing the audio signal to be denoised to obtain the spectrum feature, inputting the spectrum feature into the convolutional network model to be processed to obtainthe spectrum convolutional feature, and inputting the spectrum convolutional feature into the cyclic network model to be processed to obtain the target spectrum or the target spectrum mask; processingthe target spectrum to obtain a target audio signal; or processing the audio signal to be denoised by using the target spectrum mask to obtain a target audio signal. According to the embodiment of the invention, the neural network structure is applied to singing noise reduction of karaoke recording, so the original signal structure of the singing can be well maintained while the noisy singing issubjected to effective noise reduction, and the obvious hearing distortion after noise reduction is avoided.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to an audio noise reduction method and a training method for an audio noise reduction model. Background technique [0002] In recent years, the market size of mobile karaoke software has gradually expanded, with user groups of all ages and music levels. Especially with the popularization of intelligent terminals such as smart phones and tablet computers, it is possible for users to perform karaoke without leaving home. For example, after a user installs karaoke software on their smartphones, they can sing songs without going into a KTV. One of the main scenarios of this kind of software is song recording, that is, two audio signals of accompaniment and human voice, and finally generate one audio signal data of the synthesized work through signal processing technology. [0003] At present, when using the karaoke APP on the market to record, due to the limitation of non-prof...

AI Technical Summary

Problems solved by technology

[0004] The applicant found that this traditional noise reduction method only has a certain noise reduction effect on certain types of noise (steady state noise, etc.), and it is difficult to achieve ideal results for other more complex and changeable background noises (such as non-stationary state). Effect

In addition, the traditional singing noise reduction method is easy to introduce distortion to the original human voice signal during the noise reduction process.

Images