Abstract
This paper aims to contribute in modeling and implementation, over a system on chip SoC, of a powerful technique for phonemes recognition in continuous speech. A neural model known by its efficiency in static data recognition, named SOM for self organization map, is developed into a recurrent model to incorporate the temporal aspect in these applications. The obtained model RSOM will subsequently introduced to ensure the diversification of the genetic algorithm GA populations to expand even more the search space and optimize the obtained results. We assigned a chromosomal vision to this model in an effort to improve the information recognition rate.
Highlights
Voice recognition is complicated by the dynamic state highly variable of the speech signal
A neural model known by its power in static data recognition, named SOM for self organization map, is developed into a recurrent model to incorporate the temporal aspect in these applications
Our research focuses on the application of SOM and RSOM for phonemes recognition in TIMIT databases
Summary
Voice recognition is complicated by the dynamic state highly variable of the speech signal. A technique often used is to decompose the signal into smaller atoms under stationary states representing phonemic entities able to be treated This strategy motivates the ability to further improve the recognition score in stationary areas. A neural model known by its power in static data recognition, named SOM for self organization map, is developed into a recurrent model to incorporate the temporal aspect in these applications This idea overcomes a large adequacy between the ability robustness of the recognition tool and the speech signal to be processed. Of the training or the test stage of the RSOM model by a certain input data, appears one BMU which is illustrated by one type of RSOM map This type of map represents one chromosome of the adopted genetic algorithm GA. These individuals obtained through many iterations until a stop criterion involve populations which represent different phonemes of a continuous speech signal to be processed
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call