TIME-DEPENDENT SELF-ORGANIZING MAPS FOR SPEECH RECOGNITION

The Self-Organizing Map (SOM) algorithm [6] is used in pattern recognition tasks. In speech recognition experiments it has produced good results [9]. In this paper two modifications for SOM are proposed that unlike the original one take into account the changing of input signal. In the first a time average of a sequence of responses of one SOM is found out and this is recognised by another SOM. In the second successive input patterns are concatenated and recognised by SOM. Comparing the results to recognition system utilising the original SOM, we can improve the recognition of isolated phonemes from 10.4 % of errors to 7.0 % and 5.0 % of errors for integration model and concatenation model respectively. If the phoneme segments are also to be located the error rates decline from 9.2 % of errors to 8.2 % and 7.6 % of errors for the new methods respectively.

[1]  Jari Kangas.  Phoneme recognition using time-dependent versions of self-organizing maps , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[2]  T. Kohonen.  Self-organized formation of topographically correct feature maps , 1982 .

[3]  Geoffrey E. Hinton,et al.  Phoneme recognition using time-delay neural networks , 1989, IEEE Trans. Acoust. Speech Signal Process..

[4]  Teuvo Kohonen,et al.  Transient map method in stop consonant discrimination , 1989, EUROSPEECH.

[5]  Olli Ventä,et al.  Phonetic typewriter for Finnish and Japanese , 1988, ICASSP-88., International Conference on Acoustics, Speech, and Signal Processing.

[6]  Teuvo Kohonen,et al.  The self-organizing map , 1990 .

[7]  K.-F. Lee,et al.  Speaker-independent recognition of connected utterances using recurrent and non-recurrent neural networks , 1989, International 1989 Joint Conference on Neural Networks.