JAIST Repository >
School of Information Science >
Conference Papers >
Conference Papers >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10119/11928

Title: Speech recognition in noisy conditions based on speech separation using Non-negative Matrix Factorization
Authors: Du, Yuxuan
Akagi, Masato
Issue Date: 2014
Publisher: 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP'14)
Magazine name: 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP'14)
Start page: 429
End page: 432
Abstract: This paper proposes a speech recognition method for applications in adverse noisy environments. Speech recognition in noisy conditions is a challenging problem since speech observed in such conditions is corrupted by noise. To deal with this problem, we integrate non-negative matrix factorization (NMF) and modified restricted temporal decomposition (MRTD) into a recognition method based on the concept of “auditory scene analysis” (ASA). Experiments were conducted using 100 isolated words in 4 different noise conditions at a signal to noise ratio (SNR) of 0 dB. Experimental results showed the proposed method achieved recognition rates of 80%, which is about 50% higher than that of the method based on dynamic time warp (DTW).
Rights: This material is posted here with permission of the Research Institute of Signal Processing Japan. Yuxuan Du, Masato Akagi, 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP'14), 2014, 429-432.
URI: http://hdl.handle.net/10119/11928
Material Type: publisher
Appears in Collections:b11-1. 会議発表論文・発表資料 (Conference Papers)

Files in This Item:

File Description SizeFormat
NCSP2014_DU.pdf260KbAdobe PDFView/Open

All items in DSpace are protected by copyright, with all rights reserved.

 


Contact : Library Information Section, Japan Advanced Institute of Science and Technology