Intelligibility Enhancement of Synthetic Speech: A Review

Abstract

Current method of speech enhancement has been developed with adaptive filtering approach. The adaptive filter utilizes the least mean square algorithm for noise removal, but in LMS algorithm key parameter is step size. When step size is large speed and least mean square error is large and it is that computational cost increases to an undesirable level as the length of the impulse response increases. This paper provide a detail review on existing methodologies on enhancement on synthetic speech.