摘要
数字信号处理技术的迅速发展,为语音识别的实时实现提供了可能。采用TI公司的DSP芯片TMS320VC5409,建立一个汉语数字的语音实时识别系统。DSP的多通道缓冲串口与模数转换芯片的连接实现语音信号的采样;分别使用64k的程序和数据存储空间;系统的输出是使用TL16C550实现异步通信串口并使之与计算机RS232串口连接以观察识别结果。系统运行算法主要有字端点检测、特征提取和识别算法。实验结果表明,该系统基本能实现预期识别目标。
With the development of the digital signal processing technology, realization in real - time speech recognition system is probable. This paper tries to build a Chinese characters reeognition＇s system with theTMS320VC5409 chip. The recognition system sample speech signal by A/D chip which is connected to Multi -ehannal Buffer Serial Port in DSP. The system has two 64K - word chips as data memory and program memory. TL16C550 chip is used to extend an UART in order to communicate with PC RS232 serial port. Algorithms running on the system are Word Boundaries,ZCPA and RBF. Experiments show that aim of the recognition is realized basically.
出处
《电脑开发与应用》
2006年第4期12-14,共3页
Computer Development & Applications
基金
国家自然科学基金(No.60472094)
教育部留学回国科研基金(教外司留[2004]176号)
山西省留学回国人员科研基金(No.200224)
山西省高等学校青年学术带头人科研基金(晋教科[2004]13号文件)
山西省自然科学基金(No.20051039)
山西省高校科技研究开发项目(No.200312)
关键词
语音识别
DSP
特征提取
RBF
speech recognition, DSP, feature extraction, RBF
