Title:
音声認識方法、装置及びコンピュータプログラム
Document Type and Number:
Japanese Patent JP7282442
Kind Code:
B2
Abstract:
Provided are a voice recognition method, device, and computer-readable storage medium, said method comprising: obtaining a first loss function of a voice separation enhancement model and a second loss function of a voice recognition model (S202); performing back-propagation on the basis of the second loss function to train an intermediate model bridged between the voice separation enhancement model and the voice recognition model, to obtain a robust representation model (S204); combining the first loss function and the second loss function to obtain a target loss function (S206); performing joint training of the voice separation enhancement model, the robust representation model, and the voice recognition model on the basis of the target loss function, and ending training when a preset convergence condition is satisfied (S208).
Inventors:
王 ▲ジュン▼
林 永▲業▼
林 永▲業▼
Application Number:
JP2022520112A
Publication Date:
May 29, 2023
Filing Date:
November 12, 2020
Export Citation:
Assignee:
TENCENT TECHNOLOGY(SHENZHEN)COMPANY LIMITED
International Classes:
G10L15/06; G10L15/065; G10L15/16; G10L15/20
Domestic Patent References:
| JP2019078857A |
Foreign References:
| WO2019198265A1 | ||||
| US20180053087 | ||||
| US20190043516 |
Other References:
Max W.Y.Lam et. al.,Extract, Adapt and Recognize: an End-to-end Neural Network for Corrupted Monaural Speech Recognition,INTERSPEECH 2019,2019年09月15日,pp.2778-2782
Attorney, Agent or Firm:
Shinya Mihiro
Naoki Matsuo
Naoki Matsuo