Human Computer Interactions using Adaptive Speech-Text-Speech Conversions

In this paper, we present an adaptive method for speech-text-speech conversion and discuss about the issues relevant to the development of advanced human computer interaction. The existing STS method is modified by replacing the speech recognition method in an adaptive way. The new system helps the visually impaired people by compensating noises of an input speech using recursive least square method of Dynamic Time warping and adjusting the parameters of HMM speech model .The input speech having plurality of speech frames can be adjusted using the refinement normalisation method. These issues are especially important for robust speech recognition. This method of speech can be combined with unit selection method of speech synthesis that leads to an adaptive human computer interaction.