A Novel Noise-Robust ASR Method by Applying Partially Connected DNN Model and Mixed-Bandwidth Concept

DOI

10.2991/3ca-13.2013.46How to use a DOI?

Keywords

robust; ASR; DNN; mixed-bandwidth

Abstract

In recent years, deep neural networks achieve significant improvements in automatic speech recognition. In this paper, we propose a deep structure used for robust ASR. The model has several partially connected layers which can suppress noise in different frequency bands. In order to recognize the speech data which has been distorted by noise seriously, we try to use parts of their frequency bands with a mixed-bandwidth model. The results have shown that the partially connected network could suppress noises in different frequency bands properly. The model's phone recognition on TIMIT corpus outperforms the state-of-the-art DNN model.

Copyright

© 2013, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Lichun Fan
AU  - Hongyan Li
AU  - Dengfeng Ke
AU  - Bo Xu
PY  - 2013/04
DA  - 2013/04
TI  - A Novel Noise-Robust ASR Method by Applying Partially Connected DNN Model and Mixed-Bandwidth Concept
BT  - Proceedings of the 2nd International Symposium on Computer, Communication, Control and Automation
PB  - Atlantis Press
SP  - 182
EP  - 185
SN  - 1951-6851
UR  - https://doi.org/10.2991/3ca-13.2013.46
DO  - 10.2991/3ca-13.2013.46
ID  - Fan2013/04
ER  -