一种SmartFusion FPGA的快速语音合成系统设计

针对传统语音合成系统实时性差、硬件资源占用率高、大容量语音数据处理效率低等问题,采用高速并行运算能力的SmartFusion FPGA （Field Programmable Gate Array）为硬件开发平台,设计了一种FPGA的语音快速合成系统.首先通过移植FatFs文件系统到FPGA芯片集成的Cortex-M3硬核中,采用软硬件协同工作方式,利用Cortex-M3硬核与FatFs文件系统实现语音文件的查找和软件解析,应用FPGA硬件完成对文本文件内容读取及控制语音的播放,形成软硬件协同工作模式.实验结果表明,该语音合成系统具有语音文件处理效率高、文语转换时间短、语音播放清晰流畅,实现方便,稳定可靠.

Design of a Fast Text to Speech System Based on SmartFusion FPGA

Aiming at the problems of the traditional text to speech （TFS） systems,such as poor real-time character,high utilization of hardware resources,low efficiency of high-capacity speech data processing,it proposes the use of the SmartFusion FPGA （Field Programmable Gate Array） as the platform of hardware development.A TTS system was designed,based on FPGA.Firstly,the FatFs file system was transplanted to the Cortex-M3 hardcore of FPGA with the chip integrated.Then,hardware and software were made to work cooperatively,Cortex-M3 hardcore and FatFs file system were responsible for finding voice files and doing software analysis,and FPGA hardware was used to read the text contents and play the voice.Experimental results show that the TTS system has higher efficiency of processing voice files,shorter time of TTS conversion,and clear and fluent playback of the voice.The TTS system is stable and easy to implement.