对智能电视语音识别标准草稿中部分技术问题的思考

Thoughts on Some Technical Problems in the Standard Draft of Speech Recognition of Smart TV

语音识别技术在智能电视广泛使用颠覆了传统按键式遥控器,本文介绍了语音控制功能的工作原理,分析了语音识别技术的难点,介绍标准中最受关注的四点要求:对测试场地的要求、对模拟背景干扰噪声的要求、对识别性能指标的要求、对语音识别网络测试条件的要求。最后对标准中五个问题阐述了技术观点和理由:自由跌落高度的问题、测试距离和角度的问题、遥控器欠压状态下的指标问题、遥控器非正常使用带来的问题、语音控制功能引发隐私等方面的问题。

Speech recognition technology widely used in smart TV overturns the traditional button remote control. This paper introduces the working principle of voice control function, and analyzes the difficulties in speech recognition technology. Then it introduces the four requirements that attracts the most attention in the standard, which are requirements for test site, requirements for the background interference noise simulation, requirements for identify performance, and requirements for speech recognition network test conditions. Finally, the technical views and reasons are expounds from the perspectives of these five problem in the standard, that are the problem of free drop height, the problem of test of distance and angle, the problem of remote control under voltage state indicators, the problem of abnormal use of remote control, and the problem of voice control function such as privacy issues.