Acoustic Event Detection in Multichannel Audio Using Gated RecurrentNeural Networks with High-Resolution Spectral Features

Collection with item attached
2017
Item details URL
http://open-repository.kisti.re.kr/cube/handle/open_repository/474216.do
DOI
10.4218/etrij.17.0117.0157
Title
Acoustic Event Detection in Multichannel Audio Using Gated RecurrentNeural Networks with High-Resolution Spectral Features
Description
This work was supported by Basic Science Research Program through theNational Research Foundation of Korea (NRF) funded by the Ministry ofEducation, Rep. of Korea (NRF-2015R1D1A1A01059804). And the presentResearch has been conducted by the Research Grant of KwangwoonUniversity in 2017.
abstract
Recently, deep recurrent neural networks have achieved great success in various machine learning tasks, and have also been applied for sound event detection. The detection of temporally overlapping sound events in realistic environments is much more challenging than in monophonic detection problems. In this paper, we present an approach to improve the accuracy of polyphonic sound event detection in multichannel audio based on gated recurrent neural networks in combination with auditory spectral features. In the proposed method, human hearing perception-based spatial and spectral-domain noise-reduced harmonic features are extracted from multichannel audio and used as high-resolution spectral inputs to train gated recurrent neural networks. This provides a fast and stable convergence rate compared to long short-term memory recurrent neural networks. Our evaluation reveals that the proposed method outperforms the conventional approaches.
provenance
Made available in Cube on 2018-09-28T10:40:14Z (GMT). No. of bitstreams: 0
language
English
author
Kim, Hyoung-Gook
Kim, Jin Young
accessioned
2018-09-28T10:40:14Z
available
2018-09-28T10:40:14Z
issued
2017
citation
ETRI JOURNAL(39): 6
issn
1225-6463
uri
http://open-repository.kisti.re.kr/cube/handle/open_repository/474216.do
Funder
교육부
Funding Program
개인기초연구(교육부)
Project ID
1345270991
Jurisdiction
Rep.of Korea
Project Name
IoT-Based Next Generation Music Ecosystem
rights
openAccess
subject
Acoustic event detection
Deep recurrent neural networks
Gatedrecurrent neural network
Multichannel audio
type
article


Files in This Item

There are no attached files.