Research Article
Research on Sound Source Recognition Algorithm of Pickup Array Based on Adaptive Background Noise Removal
@INPROCEEDINGS{10.1007/978-3-030-90196-7_29, author={Chengyu Hou and Liu Can and Di Chen}, title={Research on Sound Source Recognition Algorithm of Pickup Array Based on Adaptive Background Noise Removal}, proceedings={Artificial Intelligence for Communications and Networks. Third EAI International Conference, AICON 2021, Xining, China, October 23--24, 2021, Proceedings, Part I}, proceedings_a={AICON}, year={2021}, month={11}, keywords={L-shaped pickup array Noise cancellation Sound source identification}, doi={10.1007/978-3-030-90196-7_29} }- Chengyu Hou
Liu Can
Di Chen
Year: 2021
Research on Sound Source Recognition Algorithm of Pickup Array Based on Adaptive Background Noise Removal
AICON
Springer
DOI: 10.1007/978-3-030-90196-7_29
Abstract
Nowadays, the pickup array is used in a large number of occasions, such as human voice recognition, audio conference, video conference and sound source localization. The research of sound source recognition algorithm based on pickup array has broad application prospects in the military field. The sound source recognition technology at this stage is implemented by a relatively fixed pickup array. However, due to the high requirements for the number of array elements, it faces severe environmental noise interference. Therefore, the sound source signal needs to be pre-processed before being formally processed. This paper discusses the sound source recognition algorithm based on the pickup array, which reduces the influence of environmental noise interference by preprocessing the sound source signal; realizes the target sound source recognition through feature extraction and the establishment of a recognition model. This article starts with the study of the preprocessing method of the sound source signal of the L-shaped pickup array node, and discusses an LMS noise cancellation model based on an improved variable step size. At the same time, this article identifies the target sound source signal and uses the MFCC feature extraction method. On the basis, the MFCC feature extraction method for high frequency suppression is given, and then the sound source recognition algorithm based on GMM-UBM is introduced.
