车载语音点歌系统的实现

Design and Implementation of Speech Interaction in Vehicle for Ordering Songs

基于语音识别的人机交互方式是人类进行信息交换的一种最便捷、最普遍、最自然的方式,是嵌入式车载电子产品的人机交互方面的一个发展趋势。然而,语音模板库与实际环境中输入的语音存在很大差异,因此,提高语音识别的鲁棒性成为了一个重要的课题。本文改进了谱减法,以抑制有色噪声来增强语音,提高语音交互的鲁棒性;系统采用了采用四层软件架构体系,分别为硬件驱动层、硬件抽象层、引擎层、应用层。这种分层的系统结构,可以很好的屏蔽系统各层次之间的相关性,使得系统的二次开发更加简便。

Human-Computer Interaction based on speech recognition is one of the most convenient, universal, and natural way for communication. It is also a development of Human-Computer Interaction technique applied in the embedded car kit. However, the speech information in template library is quite different from the input in actual environment. So enhancing the robustness of speech recognition system becomes an important issue in speech recognition technology. This paper improves the spectral subtraction method to enhance speech corrupted by color noisy and the robustness. The system uses a four-tiered structure including hardware driver, abstract, engine and application. This hierarchical system architecture can excellently shield the inter-influence of the four tiers and makes it more convenient and easy for the further development.