A noisy speech recognition method based on singular value decomposition

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)22_Publication in policy or professional journal

5 Scopus Citations
View graph of relations



Original languageEnglish
Pages (from-to)91-93
Journal / PublicationHuanan Ligong Daxue Xuebao/Journal of South China University of Technology (Natural Science)
Issue number1
Publication statusPublished - Jan 2001
Externally publishedYes


This paper presents a noisy speech recognition method. The speech signals for training and testing should be preprocessed by the same singular value decomposition filtering prior to extracting their features. How to decide the value of the filter's parameter is also discussed. Experimental results shows that this method can significantly raise the conventional HMM recognizer robust to noise.

Research Area(s)

  • Signal processing, Speech processing, Speech recognition