Speech enhancement and Speech recognition by improving PSNR Value

Abstract

In order to improve performance robustness of Speech enhancement and speech recognition system under difficult situations, we recommend improved Peak-Signal to Noise Ratio (PSNR).It consists of two stages .In First stage we Collect speech samples from group of people and stored in database. In the second stage, we recognize the speech from the database, whether it is present in database or not. Stochastic matching criterion is used during estimation of the super vector for a new speech signal. In the first stage, aim is speech enhancement and in the second stage aim is speech recognition with minimum error. Comparing to baseline system, speech enhancement and speech recognition system by improving PSNR provides better results.