Active and Semi-Supervised Learning in ASR: Benefits on the Acoustic and Language Models

被引:16
作者
Drugman, Thomas [1 ]
Pylkkonen, Janne [1 ]
Kneser, Reinhard [1 ]
机构
[1] Amazon, Seattle, WA 98109 USA
来源
17TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2016), VOLS 1-5: UNDERSTANDING SPEECH PROCESSING IN HUMANS AND MACHINES | 2016年
关键词
speech recognition; active learning; semi supervised training; data selection; SPEECH RECOGNITION;
D O I
10.21437/Interspeech.2016-1382
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
The goal of this paper is to simulate the benefits of jointly applying active learning (AL) and semi-supervised training (SST) in a new speech recognition application. Our data selection approach relies on confidence filtering, and its impact on both the acoustic and language models (AM and LM) is studied. While AL is known to be beneficial to AM training, we show that it also carries out substantial improvements to the LM when combined with SST. Sophisticated confidence models, on the other hand, did not prove to yield any data selection gain. Our results indicate that, while SST is crucial at the beginning of the labeling process, its gains degrade rapidly as AL is set in place. The final simulation reports that AL allows a transcription cost reduction of about 70% over random selection. Alternatively, for a fixed transcription budget, the proposed approach improves the word error rate by about 12.5% relative.
引用
收藏
页码:2318 / 2322
页数:5
相关论文
共 20 条
  • [1] [Anonymous], 1998, PROC BROADCAST NEWS
  • [2] BERGEN Z., 1997, P 5 EUR C SPEECH COM, P819
  • [3] COHN D, 1994, MACH LEARN, V15, P201, DOI 10.1007/BF00993277
  • [4] Fraga-Silva T, 2015, 16TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2015), VOLS 1-5, P3159
  • [5] SPEECH MODELING BASED ON COMMITTEE-BASED ACTIVE LEARNING
    Hamanaka, Yuzo
    Shinoda, Koichi
    Furui, Sadaoki
    Emori, Tadashi
    Koshinaka, Takafumi
    [J]. 2010 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, 2010, : 4350 - 4353
  • [6] Heigold G, 2013, INT CONF ACOUST SPEE, P8619, DOI 10.1109/ICASSP.2013.6639348
  • [7] Deep Neural Networks for Acoustic Modeling in Speech Recognition
    Hinton, Geoffrey
    Deng, Li
    Yu, Dong
    Dahl, George E.
    Mohamed, Abdel-rahman
    Jaitly, Navdeep
    Senior, Andrew
    Vanhoucke, Vincent
    Patrick Nguyen
    Sainath, Tara N.
    Kingsbury, Brian
    [J]. IEEE SIGNAL PROCESSING MAGAZINE, 2012, 29 (06) : 82 - 97
  • [8] Active Learning by Querying Informative and Representative Examples
    Huang, Sheng-Jun
    Jin, Rong
    Zhou, Zhi-Hua
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2014, 36 (10) : 1936 - 1949
  • [9] Huang Y, 2013, INTERSPEECH, P2359
  • [10] Itoh N, 2012, INT CONF ACOUST SPEE, P4133, DOI 10.1109/ICASSP.2012.6288828