Invention Grant
US08306817B2 Speech recognition with non-linear noise reduction on Mel-frequency cepstra
有权
在梅尔频率cepstra上进行非线性降噪的语音识别
- Patent Title: Speech recognition with non-linear noise reduction on Mel-frequency cepstra
- Patent Title (中): 在梅尔频率cepstra上进行非线性降噪的语音识别
-
Application No.: US11970537Application Date: 2008-01-08
-
Publication No.: US08306817B2Publication Date: 2012-11-06
- Inventor: Dong Yu , Alejandro Acero , James G. Droppo , Li Deng
- Applicant: Dong Yu , Alejandro Acero , James G. Droppo , Li Deng
- Applicant Address: US WA Redmond
- Assignee: Microsoft Corporation
- Current Assignee: Microsoft Corporation
- Current Assignee Address: US WA Redmond
- Agency: Westman, Champlin & Kelly, P.A.
- Main IPC: G10L15/00
- IPC: G10L15/00

Abstract:
In an automatic speech recognition system, a feature extractor extracts features from a speech signal, and speech is recognized by the automatic speech recognition system based on the extracted features. Noise reduction as part of the feature extractor is provided by feature enhancement in which feature-domain noise reduction in the form of Mel-frequency cepstra is provided based on the minimum means square error criterion. Specifically, the devised method takes into account the random phase between the clean speech and the mixing noise. The feature-domain noise reduction is performed in a dimension-wise fashion to the individual dimensions of the feature vectors input to the automatic speech recognition system, in order to perform environment-robust speech recognition.
Public/Granted literature
- US20090177468A1 SPEECH RECOGNITION WITH NON-LINEAR NOISE REDUCTION ON MEL-FREQUENCY CEPTRA Public/Granted day:2009-07-09
Information query