Environmental Sound Recognition Based on Residual Network and Stacking Algorithm

Haoyuan Wang, Xuemei Ren*, Zhen Zhao

*此作品的通讯作者

科研成果: 书/报告/会议事项章节会议稿件同行评审

摘要

Environmental sound recognition is one of the important tasks in the field of audio research. Because the environment is complex and there is a lot of useless sound information, the traditional methods have low recognition accuracy, which is gradually replaced by related methods of deep learning. In this paper, combined with the latest research in this field, the recognition algorithm based on residual network and stacking method is proposed. The whole is divided into two parts: a feature extractor and a classifier. The residual network is responsible for extracting features with high recognition rate and the stacking algorithm is responsible for accurate recognition. The method is applied to the representative datasets ESC-50 and UrbanSound8k. We obtain a higher accuracy and the model is more clear and simple.

源语言英语
主期刊名Proceedings of 2020 Chinese Intelligent Systems Conference - Volume II
编辑Yingmin Jia, Weicun Zhang, Yongling Fu
出版商Springer Science and Business Media Deutschland GmbH
682-690
页数9
ISBN(印刷版)9789811584572
DOI
出版状态已出版 - 2021
活动Chinese Intelligent Systems Conference, CISC 2020 - Shenzhen, 中国
期限: 24 10月 202025 10月 2020

出版系列

姓名Lecture Notes in Electrical Engineering
706 LNEE
ISSN(印刷版)1876-1100
ISSN(电子版)1876-1119

会议

会议Chinese Intelligent Systems Conference, CISC 2020
国家/地区中国
Shenzhen
时期24/10/2025/10/20

指纹

探究 'Environmental Sound Recognition Based on Residual Network and Stacking Algorithm' 的科研主题。它们共同构成独一无二的指纹。

引用此