SLHAR: A supervised learning approach for homophone ambiguity reduction from speech recognition system

Abstract

An automatic Speech to Text (STT) conversion technology has been developed for making a visual text layout of the Speech Input for advancement of Science and Technology. This technology enables people an alternative way to understand voice communication, and pursue instruction using their visual ability. The visual ability becomes more powerful than the listening ability some time more than even in remote communication, and STT conversion plays a role as an important tool in such cases. The system faces many kind of ambiguity during STT. The research focuses on the Homophone Ambiguity and with the help of the Classified Supervised learning it tries to improve it partially. In the proposed Supervised learning based Homophone Ambiguity Reduction (SLHAR), a large dataset are taken as homophones and Homophone Sets are assembled by Hierarchical Clustering Method. The proposed system communicates with the user in case of Homophones and converts them to the text format.