Speech Recognition from PSD using Neural Network

in this paper we present a system for speech recognition using neural network from obtained data of power spectral density peaks. For this work, a small size vocabulary containing the word "yes" and "no" is chosen. Spectrum features is extracted from estimated power spectrum by Autoregressive parametric method in each frame of speech signal. And it is given to feed forward Back propagation neural network with gradient descent with adaptive learning rate training algorithm. Network is trained for classification to two classes.

[1]  Khaled Benkrid,et al.  Proceedings of the International MultiConference of Engineers and Computer Scientists, Vol II, IMECS 2009 , 2009 .

[2]  Aki Härmä,et al.  Frequency-warped autoregressive modeling and filtering , 2001 .

[3]  S. L. Marple,et al.  A tutorial overview of modern spectral estimation , 1989, International Conference on Acoustics, Speech, and Signal Processing,.

[4]  Jun Yang,et al.  Isolated speech recognition using artificial neural networks , 2001, 2001 Conference Proceedings of the 23rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[5]  Steven Kay,et al.  Modern Spectral Estimation: Theory and Application , 1988 .

[6]  Chee Peng Lim,et al.  Speech recognition using artificial neural networks , 2000, Proceedings of the First International Conference on Web Information Systems Engineering.

[7]  Hossein Tajik,et al.  A Novel Approach to Very Fast and Noise Robust,Isolated Word Speech Recognition , 2006, 18th International Conference on Pattern Recognition (ICPR'06).