Universal Languages Recognition Algorithm: Adapting to Various Liguistics

This project proposes a new intelligent and efficient voice recognition algorithm that not only guarantees recognition of speech in any language but, also aims to remove the redundant noise from the external environment in day to day life, in a cost effective manner. It´s another salient feature is its ´Customization´ which allows user to lock on to the desired speaker and listener for it adaptively selects a set of optimal data based on parameters it had been learning over the time. Our study indicates that under the traditional voice recognition algorithm, the end-to-end data works on efficient use of a pre-made dictionary rather than creating its own. The proposed algorithm adapts itself to the speaker and listeners frequency variations and then finally locks on to both listener and speaker. Thus, removing the redundant noise. The algorithm made is called ´speak smart´ which eventually creates a ´spectrum bank´ and then finally amalgates itself with a couple of electronic software to give it a finished hardware touch, in practical sense. This paper therefore aims to introduce the new algorithm which is effective not only in speech processing application in different languages but, is also useful in Robotics application.