Fuzzy partition models and their incremental training for continuous speech recognition

スポンサーリンク

概要

• 論文の詳細を見る

• This paper describes the application of Fuzzy Partition Models(FPMs)and their incremental training to continuous speech recognition. FPMs are neural networks with multiple input-output units. Since the outputs are non-negative and their sum is one, they can be regarded as the probabilities of recognizing input speech phonemes. Automatic incremental training is developed using the Viterbi alignment to adapt FPMs to continuous speech. The FPMs are retrained automatically by using speech data segmented by the Viterbi alignment. We combined FPMs with an LR parser(FPM-LR)and carried out experiments in continuous speech recognition. The recognition rate of the FPM-LR was higher than of a Time-Delay Neural Network-LR(TDNN-LR). Automatic incremental training was more effective with FPMs than with TDNNs.

• 社団法人日本音響学会の論文

著者

• Kato Y

  Kyoto Univ. Kyoto Jpn

• Sugiyama Masahide

  Atr Interpreting Telephony Research Laboratories

• Kato Yoshinaga

  ATR Interpreting Telephony Research Laboratories

• Sugiyama M

  Atr Interpreting Telephony Research Laboratories

関連論文

• A practical method for estimating the headway distribution based on the observed data of the number of flowing vehicles : Gamma distribution model
• Fuzzy partition models and their incremental training for continuous speech recognition
• An optimization technique for speaker mapping neural networks using minimal classification error criterion

スポンサーリンク