Research And Implementation Of Speech Synthesis Technology Based On Embedded Platform

ABSTRACT:The research subject of this thesis originates from the car terminal project team of WatchData. With ITS showing huge market potential, WatchData considers ITS services as a new growth point of its business, and there's no doubt that intelligent terminal is an important part of ITS. Therefore, WatchData starts research on intelligent car terminal of a new generation. Intelligent terminal of WatchData is based on Android system, and its bottom hardware platform is S3C6410 produced by Samsung. The embedded operation system on the hardware layer adopts Linux open source core.This thesis is engaged in research on TTS service in the subfunction of man-machine interaction in a car terminal.This thesis starts its research mainly from embedded-platform-based speech synthesis technology. First, it gives a summary and overview for various existing voice technologies, and then two appropriate embedded speech synthesis algorithms are summed up after analysis of open source application:one is diphone-based concatenative synthesis with its open source code named Flite, and this algorithm reduces the volume of speech database; the other one is based on formant synthesis with its open source code named eSpeak, and this algorithm aims at simulating sound production process of human beings. Finally, according to comparison between the two abovementioned algorithms, this thesis selects eSpeak as its target software and reconstructs the open source code and integrates its existing system. Then this software is transplanted to Android platform and the application of speech synthesis service of Chinese language in a car terminal of WatchData is achieved finally.