嵌入式实时英语语音识别系统的设计与研究

The design and research of embedded real time English speech recognition system

针对本次英语语音识别系统的设计之中,应用嵌入式技术,可以应用嵌入式DSP语音芯片,以连续隐含马尔可夫模型为基本识别框架,优化设计英语语音识别系统。结果证实,设计出基于嵌入式的实时英语语音识别系统,可以在实际应用中实时识别出不同的英语语音,而且也能够提升系统在应用中对于英语语音的识别率,提升29.0%,对于英语语音的识别速度也快,发挥积极应用效益。研究结论表明,设计嵌入式的实时英语语音识别系统,具有可行性,不仅可以提升系统识别英语语音的速度,也发挥积极的应用效益,可以在实践中推广该系统设计方法。

For the English speech recognition system design, the application of embedded technology, embedded application voice DSP chip, the continuous hidden Markov model as the basic recognition framework for optimal design of English speech recognition system. The results confirm that the design based on embedded real-time English speech recognition system, in the practical application of real-time identification of the different English pronunciation, but also can enhance system in the application for the play English speech recognition rate, improve 29.0%, for English speech recognition speed positive effect in application. Research results show that the design of embedded real- speech recognition system is feasible, not only can lifting is also fast, time English system English speech recognition rate, also play positive benefits for the application and can be popularized in practice the system design method.