Multi-channel speech enhancement method adopting microphone array pickup

Abstract

The invention discloses a multi-channel speech enhancement method adopting microphone array pickup. The method comprises the following steps: performing multi-channel pickup on speech by using a plurality of single-channel microphones and performing preprocessing; based on the selection of a reference microphone, reducing the signal difference between channels by adopting two feature extraction modes of NCC and MCS and combining an average mode, and designing a filter corresponding to the reference microphone; carrying out NCC and MCS feature extraction on other single-channel microphones, and designing filters in one-to-one correspondence with other single-channel reference microphones; and the enhanced voice signals of all the single-channel microphones are combined and averaged to obtain a final enhanced voice signal. According to the multi-channel speech enhancement method provided by the invention, the signal difference between the channels is reduced by adopting the NCC and MCS feature extraction and united averaging mode, and a good noise reduction effect is achieved on the noisy speech of the far and near fields in the indoor space.

Description

technical field [0001] The invention relates to the technical field of speech noise reduction, in particular to a multi-channel speech enhancement method using a microphone array to pick up sound. Background technique [0002] In real production and living scenarios, all kinds of complex noise interference will seriously affect the communication quality of people using voice equipment or the recognition performance of voice recognition systems. It is necessary to improve the clarity and intelligibility of the voice through the front-end processing of the voice signal in order to improve the system performance. Generally speaking, multiple microphones are used to form a certain form of array structure for multi-channel sound pickup. In addition to the time domain characteristics and frequency domain characteristics of the speech signal, more spatial feature information can be obtained than a single microphone pickup, so that Carry out higher-dimensional speech enhancement al...

AI Technical Summary

Problems solved by technology

[0003] 1. The algorithm is based on some mathematical assumptions and noise spectrum estimation. These assumptions and estimates only conform to the characteristics of the scene to a certain extent, which makes the noise reduction performance of the system limited;

[0004] 2. This type of method cannot enhance the useful voice signal outside the beam, and cannot meet the requirements of full-scene sound pickup;

[0005] 3. Due to the different distance between the speaker and the microphone in the real scene, it is difficult to ensure the uniform distribution of the characteristic energy of the speaker's voice signal. In a sentence, some segments have loud voices, and some segments have low voices. The system voice enhancement effect is unstable

Images