Teacher Speech Recognition in Smart Classrooms of Colleges and Universities Based on Audio-To-Text Conversion and Keyword Matching

Automatic identification of key words in teachers' audio in smart classrooms of colleges and universities and automatic identification of teachers' video behaviors are a new attempt in the evaluation of teaching quality of college teachers. Based on the massive audio and video data generated by smart classrooms, a continuously updated keyword library is first constructed, and then the teacher's audio is converted into text using a private cloud and a separately developed audio-to-text model. Finally, a text keyword matching model is constructed using Python3.9 to realize the identification of key words such as teachers' ethics, classroom management, course-based ideological and political education, and teaching emphasis and difficulty in the audio of college teachers' teaching. Based on this model and the automatic identification model of teachers' video behaviors, an attempt is made to automatically, fully, and objectively evaluate the teaching quality of college teachers in smart classrooms, and to solve the difficulties and pain points in the traditional evaluation of the teaching quality of college smart classroom teachers.