This work is subjected to copyright. All rights are reserved whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illusions, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication of parts thereof is permitted only under the provision of the copyright law 1965, in its current version, and permission of use must always be obtained from CSC Publishers. SPIJ Journal is a part of CSC Publishers

Artificial neural networks have been applied successfully in many static systems but present some weaknesses if patterns involve a temporal component. Let’s note for example in speech recognition or contextual information, where different of the time interval, is crucial for comprehension. Speech, being a temporal form of sensory input, is a natural candidate for investigating temporal coding in neural networks. It is only through comprehension of the temporal relationship between different sounds which make up a spoken word or sentence that speech becomes intelligible. In fact we present in this paper presents three variants of self-organizing maps (SOM), the Leaky Integrators Neurons (LIN), the Spiking_SOM (SSOM) and the recurrent Spiking_SOM (RSSOM) models. The proposed variants is like the basic SOM, however it represents the characteristic to modify the learning function and the choice of the best matching unit (BMU). The case study of the proposed SOM variants is word recognition in continuous speech and speaker independent. The proposed SOM variants show good robustness and high word recognition rates.