| Issue |
MATEC Web Conf.
Volume 173, 2018
2018 International Conference on Smart Materials, Intelligent Manufacturing and Automation (SMIMA 2018)
|
|
|---|---|---|
| Article Number | 03059 | |
| Number of page(s) | 5 | |
| Section | Digital Signal and Image Processing | |
| DOI | https://doi.org/10.1051/matecconf/201817303059 | |
| Published online | 19 June 2018 | |
Environmental sound classification based on feature fusion
1
Communication University of China, Department of Computer Science, Beijing, China
2
Communication University of China, Department of Computer Science, Beijing, China
3
Communication University of China, Department of Computer Science, Beijing, China
4
Communication University of China, Department of Computer Science, Beijing, China
* Corresponding author: zhaohuimin_f@163.com
With deep great breakthroughs of deep learning in the field of computer vision, the field of audio recognition has gradually introduced deep learning methods and achieved excellent results. These results are mainly for speech and music recognition research, and there is very little research on environmental sound classification. In recent years, people have begun to expand the research object of deep learning to the environmental sound, and achieved certain results. In this paper, we use ESC-50 as our test set, based on the SoundNet network and EnvNet network to propose a feature fusion method[1]. After the features extracted by SoundNet and EnvNet were merged, they were classified using fusion features. Experimental results show that this method has better classification accuracy for the recognition of environmental sounds than using either of the two networks separately for classification.
© The Authors, published by EDP Sciences, 2018
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.