随着互联网和人工智能技术的发展，也基于之前疫情的大时代背景，无人监考的智慧考场逐渐走进人们的视野，并且得到了初步应用。智慧考场是基于大数据和人工智能技术，充分的结合行为分析、语音识别等技术，对考生行为进行智能分析，从而实现无人监考。目前对于学生人脸表情图像信息的研究较多，但对于考场广播音频信息的分析工作却较为稀少，获取考场广播并实现对音频信息的自动识别将是本文的研究重点。目前，国内外语音识别技术发展迅猛，本文将利用科大讯飞的SDK接口实现语音识别技术的简单应用，同时重点使用MSAR语音识别技术，于GitHub平台下载开源代码调试并训练模型。将训练所得模型用于考场广播测试，之后结合有关的考场监考规则和广播系统建设原则，进一步得出智慧考场广播系统建设的参考意见。

With the development of the Internet and artificial intelligence technology, and based on the background of the previous epidemic, the intelligent examination room without invigilator has gradually entered people 's vision and has been preliminarily applied. Intelligent examination room is based on big data and artificial intelligence technology, fully combined with behavior analysis, speech recognition and other technologies, intelligent analysis of candidates ' behavior, so as to realize unmanned invigilation. There are many researches on students ' facial expression image information, but the analysis of the audio information of the examination room broadcast is relatively rare. Obtaining the examination room broadcast and realizing the automatic recognition of the audio information will be the focus of this paper. At present, speech recognition technology is developing rapidly at home and abroad. This paper will use the SDK interface of IFLYTEK to realize the simple application of speech recognition technology, and focus on the use of MSAR speech recognition technology, download the open source code debugging and training model on the GitHub platform. The training model is used for the examination room broadcast test, and then combined with the relevant examination room invigilation rules and broadcasting system construction principles, the reference opinions for the construction of the intelligent examination room broadcasting system are further obtained.