摘要
传统的基于鼠标和键盘的人机交互方式已无法满足需要,自动语音识别技术正是解决这个问题的重要方向。阐述了一种基于短时能量和过零率的双门限端点检测方法,和基于HMM的自动语音识别技术原理以及具体工程应用实例:智能媒资检索系统V1.0和气象档案管理系统V1.0,该系统解决了传统的媒资管理系统出现的问题,提高了媒资文件的管理效率、检索率和利用率。
The traditional human - computer interaction way have been not able to meet the need, which is based on mouse and keyboard. The automatic speech recognition technology is the important direction to solve the problem. A kind of double - threshold voice activity detection method is expounded, that is short - time energy and the zero rate. Also the prin- ciple of automatic speech recognition based on HMM and two specific engineering application examples is introduced in this paper, such as intelligent media -asset retrieval system vl. 0, easy media -asset management system V1.0. They can solve the problems in the traditional media asset management system and improve the efficiency, retrieval rate and utilization rate of the media asset management system.
出处
《电声技术》
2015年第8期58-62,共5页
Audio Engineering
关键词
端点检测
自动语音识别技术
智能媒资检索系统V1.0
气象档案管理系统V1.0
voice activity detection
automatic speech recognition
intelligent media - asset retrieval system V1.0
easy media - asset management system V1.0
