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

facebooktwittermailshare

Spectral feature mapping with mimic loss for robust speech recognition

Abstract: 

For the task of speech enhancement, local learning objectives are agnostic to phonetic structures helpful for speech recognition. We propose to add a global criterion to ensure de-noised speech is useful for downstream tasks like ASR. We first train a spectral classifier on clean speech to predict senone labels. Then, the spectral classifier is joined with our speech enhancer as a noisy speech recognizer. This model is taught to imitate the output of the spectral classifier alone on clean speech. This \textit{mimic loss} is combined with the traditional local criterion to train the speech enhancer to produce de-noised speech. Feeding the de-noised speech to an off-the-shelf Kaldi training recipe for the CHiME-2 corpus shows significant improvements in WER.

up
0 users have voted:

Paper Details

Authors:
Peter Plantinga, Adam Stiff, Eric Fosler-Lussier
Submitted On:
16 April 2018 - 3:17am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Deblin Bagchi
Paper Code:
3472
Document Year:
2018
Cite

Document Files

icassp-2018-poster_deblin.pdf

(57 downloads)

Subscribe

[1] Peter Plantinga, Adam Stiff, Eric Fosler-Lussier , "Spectral feature mapping with mimic loss for robust speech recognition", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2909. Accessed: Aug. 21, 2018.
@article{2909-18,
url = {http://sigport.org/2909},
author = {Peter Plantinga; Adam Stiff; Eric Fosler-Lussier },
publisher = {IEEE SigPort},
title = {Spectral feature mapping with mimic loss for robust speech recognition},
year = {2018} }
TY - EJOUR
T1 - Spectral feature mapping with mimic loss for robust speech recognition
AU - Peter Plantinga; Adam Stiff; Eric Fosler-Lussier
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2909
ER -
Peter Plantinga, Adam Stiff, Eric Fosler-Lussier . (2018). Spectral feature mapping with mimic loss for robust speech recognition. IEEE SigPort. http://sigport.org/2909
Peter Plantinga, Adam Stiff, Eric Fosler-Lussier , 2018. Spectral feature mapping with mimic loss for robust speech recognition. Available at: http://sigport.org/2909.
Peter Plantinga, Adam Stiff, Eric Fosler-Lussier . (2018). "Spectral feature mapping with mimic loss for robust speech recognition." Web.
1. Peter Plantinga, Adam Stiff, Eric Fosler-Lussier . Spectral feature mapping with mimic loss for robust speech recognition [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2909