摘要
语音控制是当前最具发展前景的新兴技术之一,目前已广泛应用于车载终端、智能机器人等场合。对语音识别在玩具车上的应用进行了深入研究。基于高度集成的DSP芯片进行语音控制系统的设计,使用了双硅麦减噪,能够获取到清晰的音频数据。根据DTW算法和动态模板匹配,对输入语音信号进行了端点检测,对特定人的孤立词进行了识别。当多个语意近似词汇无法准确识别时,提供选择处理,使语音引擎对音质、音色的区别进行了灵活的处理,提高了识别率和用户体验感受。该语音识别在正常生活环境中进行了测试,能适应日常生活或工作环境。研究结果表明,此方案和算法具有准确的识别效果。
The voice control is one of the most promising emerging technologies, and it currently has been widely used in auto motive terminals, intelligent robot, etc. The application of speech recognition in the toy car is studied. By DSP chips for voice con- trol system based on a highly integrated design, use the double silicon wheat noise reduction to achieve the audio data clearly. Based on DTW algorithm and dynamic template matching, the input speech in signal endpoint is detected, and someone's isolated word is recognized. When multiple semantic approximation vocabulary is failed to be correctly identified, provide selection processing, make the speech engine timbre and tone color different for the flexible processing, improve the recognition rate and the user experience. The speech recognition is tested in the normal life environment, and it can adapt to application in daily life or work environment. The results show that the scheme and algorithm has accurate recognition effect.
出处
《新技术新工艺》
2015年第9期84-86,共3页
New Technology & New Process
基金
中国南方电网有限责任公司科技项目(030600KK52140058)
关键词
语音控制
人机交互
智能系统
研究
voice control, human computer interaction, intelligent system, research