Loading...

Noise reduction algorithm for robust speech recognition using MLP neural network

Ghaemmaghami, M. P ; Sharif University of Technology

870 Viewed
  1. Type of Document: Article
  2. DOI: 10.1109/PACIIA.2009.5406411
  3. Abstract:
  4. We propose an efficient and effective nonlinear feature domain noise suppression algorithm, motivated by the minimum mean square error (MMSE) optimization criterion. Multi Layer Perceptron (MLP) neural network in the log spectral domain minimizes the difference between noisy and clean speech. By using this method as a pre-processing stage of a speech recognition system, the recognition rate in noisy environments is improved. We can extend the application of the system to different environments with different noises without re-training it. We need only to train the preprocessing stage with a small portion ofnoisy data which is created by artificially adding different types of noises from the NOISEX-92 database to the TIMIT speech database. Experimental results show that the proposed method can achieve significant improvement ofrecognition rates. ©2009 IEEE
  5. Keywords:
  6. Log spectral ; MLP neural network ; Robust speech recognition ; Clean speech ; Log-spectral domain ; Minimum mean-square error ; Multi layer perceptron ; Noise reduction algorithms ; Noise suppression algorithm ; Noisy environment ; Nonlinear features ; Optimization criteria ; Pre-processing ; Recognition rates ; Speech database ; Speech recognition systems ; Computational efficiency ; Industrial applications ; Neural networks ; Noise abatement ; Optimization ; Speech recognition
  7. Source: PACIIA 2009 - 2009 2nd Asia-Pacific Conference on Computational Intelligence and Industrial Applications, 28 November 2009 through 29 November 2009 ; Volume 1 , 2009 , Pages 377-380 ; 9781424446070 (ISBN)
  8. URL: http://ieeexplore.ieee.org/document/5406411