A Novel Approach to Noise Reduction and Real-Time Enhancement of Speech Synthesis

Abstract

This paper is proposed for an explicit synthetic speech on LPC speech synthesis by discovering a novel parameter for speech enhancement. We present the NSSA(Natural Synthetic Speech Approach) to converge the same input utterances by different speakers to a new common parameter. The analysis of phonemes is beingused in order to make a much more natural and clear synthesize with minimum noise. In this paper, we enhanced the real time speech synthesis with a novel approachfor advance noise reduction and high fidelity of the synthesizer system. According to this approach, many conventional applications such as text-to-speech synthesiscan be used more efficiently. Therefore we implement the Linear Predictive Coding with Levinson-Durbin algorithm for synthesizer via TMS320C6713 DSK with RTDXmodule for testing and optimizing the results to create the best enjoyable synthetic voice.