Abstract: The paper brings forward an idea of integrated real-time processing of Chinese text-to-speech conversion and speech coding & decoding on DSP. Further more, this paper introduces both the hardware and software realization of integrated real-time processing on DSP chip. Integrative real-time processing based on DSP has the following characteristics: (1) a Chinese text-to-speech system for arbitrary text has been realized on a minimum single DSP system with capacity of no more than 1Mbytes;(2) flexible customization speech function is provided to improve naturalness of synthetic speech;(3) according to the communication channel capability, it has achieved flexible speech communication with speech synthesis or speech coding;(4) on-line reconstruction for integrative real-time processing based on DSP;(5) modularized structure for DSP software, it includes command-expositive engine, synthesis engine, coding engine etc, based on universal asynchronous serial port of RS232.

摘要： 提出了汉语文语转换和语音编解码的基于DSP的实时综合处理的思想。介绍了实时综合处理的基于DSP的硬件与软件实现。文章介绍的基于DSP的实时综合处理具有以下几个方面的特点:(1)在单片DSP和不超过1MB的DSP最小系统上实现了一个任意文本的汉语文语转换系统;(2)为提高语音合成的自然度,提供了灵活的语音定制功能;(3)根据通信信道容量大小,实现可灵活选择的采用语音合成方式的语音通信和语音编码方式的语音通信;(4) 基于DSP的实时综合处理的在线重构;(5)基于RS232的通用异步串口的包括命令解释引擎、合成引擎、编码引擎等的模块化DSP软件结构。

关键词: 可编程通用DSP, 语音合成, 语音编码, 语音处理

CLC Number: 

• TN912.3