In this paper, an attempt is made to study speech recognition system with classifiers Self Organized Maps, Multilayer Perceptron, Radial Basis Function neural networks, Modular neural network, Time Lagged neural network and to develop SOM based speech recognition system. The training parameters varies as the classifier varies. In this paper a novel SOM - based speech recognition system is developed to find the nearest neighbour classification by using Euclidean and weighted Euclidean weighted distances. In order to find out the outlier of the speakers, Nearest neighbour classification by using Euclidean and Weighted Euclidean distances are developed. The promising results are obtained with good degree of accuracy.

1. T. Deselaers, G. Heigold, and H. Ney. Speech Recognition with State-based Nearest Neighbour Classifiers. In Interspeech, pages 2093-2096, Antwerp, Belgium, August 2007.
2. CHEN, O. ABDULLA, W.H., SALCIC, Z. 'Performance Evaluation of Different Front-End Processing for Speech Recognition Systems', Technical Report, SoE-621, 2005, pp1-36
3. C.J.Wellekens,"Introduction to Speech Recognition Using Neural Networks", Proceedings of ESANN-1998, pp.227-236, Bruges
4. Speech recognition Applications by Steve Green Dec 05/Jan 06.
5. Jihene El Malik, (1998). “Kohonen Clustering Networks For Use In Arabic Word Recognition System.” Sciences Faculty of Monastir, Route de Kairouan, 14-16 December.
6. C.M.Bishop, Neural Networks for pattern recognition, oxford university press, 1995.
7. K. Haese, “Self-organizing feature maps with self-adjusting learning parameters”, IEEE Transactions on Neural Networks, vol. 9, pp. 1270–1278, 1998.
8. Ben Gold and Nelson Morgan (2007) Speech and Audio Signal Processing, Wiley India Edition, New Delhi.
9. Yegnanarayana, B, (2006), Artificial neural networks Prentice-Hall of India, New Delhi.

Pitch Intensity Mel-frequency cepstral coefficient Linear predictive coefficient Recognition rate