Committee-Based Active Learning for Speech Recognition

この論文にアクセスする

この論文をさがす

著者

抄録

We propose a committee-based method of active learning for large vocabulary continuous speech recognition. Multiple recognizers are trained in this approach, and the recognition results obtained from these are used for selecting utterances. Those utterances whose recognition results differ the most among recognizers are selected and transcribed. Progressive alignment and voting entropy are used to measure the degree of disagreement among recognizers on the recognition result. Our method was evaluated by using 191-hour speech data in the Corpus of Spontaneous Japanese. It proved to be significantly better than random selection. It only required 63h of data to achieve a word accuracy of 74%, while standard training (i.e., random selection) required 103h of data. It also proved to be significantly better than conventional uncertainty sampling using word posterior probabilities.

収録刊行物

  • IEICE transactions on information and systems

    IEICE transactions on information and systems 94(10), 2015-2023, 2011-10-01

    一般社団法人 電子情報通信学会

参考文献:  15件中 1-15件 を表示

被引用文献:  1件中 1-1件 を表示

各種コード

  • NII論文ID(NAID)
    10030193524
  • NII書誌ID(NCID)
    AA10826272
  • 本文言語コード
    ENG
  • 資料種別
    ART
  • ISSN
    09168532
  • データ提供元
    CJP書誌  CJP引用  J-STAGE 
ページトップへ