Y
Yoshikazu Yamaguchi
Researcher at Nippon Telegraph and Telephone
Publications - 60
Citations - 310
Yoshikazu Yamaguchi is an academic researcher from Nippon Telegraph and Telephone. The author has contributed to research in topics: Acoustic model & Feature (machine learning). The author has an hindex of 9, co-authored 59 publications receiving 278 citations.
Papers
More filters
Proceedings ArticleDOI
Domain adaptation of DNN acoustic models using knowledge distillation
TL;DR: A novel domain adaptation method for DNN acoustic models based on the knowledge distillation framework that can effectively avoid the overfitting of convolutional neural network based acoustic models and yield lower error rates than conventional adaptation methods.
Patent
Scheme for model adaptation in pattern recognition based on Taylor expansion
TL;DR: In this paper, a model adaptation scheme determines a change in a parameter expressing a condition of pattern recognition and probabilistic model training between an initial condition and a current condition at a time of actual recognition.
Proceedings ArticleDOI
Encoder Transfer for Attention-based Acoustic-to-word Speech Recognition.
Sei Ueno,Takafumi Moriya,Masato Mimura,Shinsuke Sakai,Yusuke Shinohara,Yoshikazu Yamaguchi,Yushi Aono,Tatsuya Kawahara +7 more
TL;DR: Domain adaptation based on transfer learning with layer freezing is proposed for adaptation of the latent linguistic capability of the decoder to the target domain and the models trained with the proposed method achieved better accuracy than the baseline models.
Proceedings ArticleDOI
Progressive Neural Network-based Knowledge Transfer in Acoustic Models
Takafumi Moriya,Ryo Masumura,Taichi Asami,Yusuke Shinohara,Marc Delcroix,Yoshikazu Yamaguchi,Yushi Aono +6 more
TL;DR: A new architecture based on progressive neural networks (PNN) that can transfer knowledge; it does not forget and can well utilize pre-trained knowledge is proposed and an enhanced PNN that uses feature augmentation to better leverage pre- trained knowledge is introduced.
Patent
Speech segment speaker classification device and method therefore, speech recognition device using the same and method therefore, program and recording medium
TL;DR: In this article, the authors proposed a speech segment speaker classification device that includes; a sound volume voice period dividing section, a feature value analysis section, representative feature value extracting section, segment classification section, and a segment integration section.