Voice2SQL: A Hybrid Intelligent System for the Recovery of Information from Databases by Means of Spoken Natural Language

In this article, it is proposed a model of recovery of information from a database through queries in Natural Language using the voice as input vector. This model incorporates a Hybrid Intelligent System based on Genetic Algorithms and a Self-Organizing Map of Kohonen (SOM) to recognize the present phonemes in a word through time, which allows us to remake up a word with speaker independence. Besides, it is proposed the use of a compiler with grammar of type 2 according to the Chomsky Hierarchy to support the syntactic and semantic structure in Spanish language. Our experiments suggest that the voice and Natural Language improve notably the Human-Computer interaction regarding with traditional methods, also that it is possible to achieve a more exact recognition of words by using neural networks SOM regarding to previous approaches as Hidden Markov Models (HMM).