节点文献
火电厂语音报警系统中的动态文语转换方法
Text-to-speech conversion method of dynamic speech alert system for power plant
【摘要】 针对火电厂实时动态语音报警系统的特点,提出了基于动态文语转换技术的火电厂实时动态语音报警系统的实现框架.阐述了实时、动态文语转换子系统的各组成部分,研究了基于基音同步叠加算法(PSOLA)的实时动态文语转换方法,讨论了采用组件对象模型(COM)技术进行包括语音合成数据库、文本预处理、韵律处理和语音波形的生成在内的程序实现的要点.分析测试的结果表明,采用这种方法合成的语音质量高、实时性好,完全可以应用到包括火电厂实时动态语音报警系统在内的其他实时语音处理场合中.
【Abstract】 After analyzing the features of a real-time dynamic speech alert system for power plant, each part of the text-to-speech subsystem was described.A real-time dynamic text-to-speech conversion method based on pitch synchronous overlap add(PSOLA) algorithm was proposed,and the key points for programming using component object model(COM) were discussed,including corpus for speech synthesis,pre-handling of Chinese texts,dealing with rhythm,and generation of speech waveforms.The preliminary hearing test results show that this system has low computational cost,can synthesize high quality speech, and can be effectively used in real-time dynamic speech alert system for power plant and other real-time speech processing.
【Key words】 power plant; text-to-speech conversion technology; pitch synchronous overlap add algorithm; speech alert;
- 【文献出处】 浙江大学学报(工学版) ,Journal of Zhejiang University(Engineering Science) , 编辑部邮箱 ,2007年12期
- 【分类号】TM76
- 【被引频次】4
- 【下载频次】95