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

facebooktwittermailshare

Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders

Abstract: 

We present Mockingjay as a new speech representation learning approach, where bidirectional Transformer encoders are pre-trained on a large amount of unlabeled speech. Previous speech representation methods learn through conditioning on past frames and predicting information about future frames. Whereas Mockingjay is designed to predict the current frame through jointly conditioning on both past and future contexts. The Mockingjay representation improves performance for a wide range of downstream tasks, including phoneme classification, speaker recognition, and sentiment classification on spoken content, while outperforming other approaches. Mockingjay is empirically powerful and can be fine-tuned with downstream models, with only 2 epochs we further improve performance dramatically. In a low resource setting with only 0.1% of labeled data, we outperform the result of Mel-features that uses all 100% labeled data.

up
0 users have voted:

Paper Details

Authors:
Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee
Submitted On:
15 May 2020 - 10:18pm
Short Link:
Type:
Presentation Slides
Event:
Presenter's Name:
Andy T. Liu
Paper Code:
SPE-L13.2
Document Year:
2020
Cite

Document Files

Presentation Slides

(25)

Subscribe

[1] Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee, "Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders", IEEE SigPort, 2020. [Online]. Available: http://sigport.org/5364. Accessed: Aug. 12, 2020.
@article{5364-20,
url = {http://sigport.org/5364},
author = {Andy T. Liu; Shu-wen Yang; Po-Han Chi; Po-chun Hsu; Hung-yi Lee },
publisher = {IEEE SigPort},
title = {Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders},
year = {2020} }
TY - EJOUR
T1 - Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders
AU - Andy T. Liu; Shu-wen Yang; Po-Han Chi; Po-chun Hsu; Hung-yi Lee
PY - 2020
PB - IEEE SigPort
UR - http://sigport.org/5364
ER -
Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee. (2020). Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders. IEEE SigPort. http://sigport.org/5364
Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee, 2020. Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders. Available at: http://sigport.org/5364.
Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee. (2020). "Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders." Web.
1. Andy T. Liu, Shu-wen Yang, Po-Han Chi, Po-chun Hsu, Hung-yi Lee. Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders [Internet]. IEEE SigPort; 2020. Available from : http://sigport.org/5364