Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM

Abstract: 

In automatic speech recognition (ASR), error correction after the initial search stage is a commonly used technique to improve performance. Whilst completely automatic error correction, such as full second pass rescoring using complex language models, is widely used, directed error correction, where the error locations are manually given, is of great interest in many scenarios. Previous works on directed error correction usually uses the error location information to change search space with original ASR models. In this paper, a novel deep learning based score combination approach is proposed for directed error correction. Here, a bi-directional LSTM (BLSTM) language model is trained to estimate unnormalized sentence completion scores. These completion scores are then combined with the confusion network scores from the initial search stage for hypothesis rescoring. Experiments showed that the BLSTM based language model achieved better results not only than simpler models such as bi-directional n-gram or LSTM, but also better than human prediction. In a real world Chinese ASR task, it was also shown that the proposed approach significantly outperformed the approach of choosing the second best hypothesis in the error sausages of confusion networks.

poster.pdf

PDF icon poster.pdf (691 downloads)
up
0 users have voted:

Paper Details

Authors:
Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu
Submitted On:
18 October 2016 - 1:03pm
Short Link:
Type:
Poster
Event:
Presenter's Name:
Da Zheng
Paper Code:
122
Document Year:
2016
Cite

Document Files

poster.pdf

(691 downloads)

Subscribe

[1] Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu, "Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM", IEEE SigPort, 2016. [Online]. Available: http://sigport.org/1259. Accessed: Nov. 23, 2017.
@article{1259-16,
url = {http://sigport.org/1259},
author = {Da Zheng; Zhehuai Chen; Yue Wu; Kai Yu },
publisher = {IEEE SigPort},
title = {Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM},
year = {2016} }
TY - EJOUR
T1 - Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM
AU - Da Zheng; Zhehuai Chen; Yue Wu; Kai Yu
PY - 2016
PB - IEEE SigPort
UR - http://sigport.org/1259
ER -
Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu. (2016). Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM. IEEE SigPort. http://sigport.org/1259
Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu, 2016. Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM. Available at: http://sigport.org/1259.
Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu. (2016). "Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM." Web.
1. Da Zheng, Zhehuai Chen, Yue Wu, Kai Yu. Directed Automatic Speech Transcription Error Correction Using Bidirectional LSTM [Internet]. IEEE SigPort; 2016. Available from : http://sigport.org/1259