Abstract: With the development of intelligent education, the analysis of classroom teaching behavior is no longer limited to traditional observation and feedback methods. Artificial intelligence technology, especially speech recognition and facial expression analysis, provides more accurate and comprehensive evaluation tools. This article explores how to improve the accuracy and effectiveness of classroom interaction by combining recognition speech data and facial analysis data, and provides support for teachers to optimize teaching strategies. The research methods mainly include data acquisition techniques, deep learning models, and multimodal data fusion analysis. Research has shown that comprehensive analysis of speech and facial data can more comprehensively evaluate students' learning status, classroom atmosphere, and teachers'teaching performance, providing real-time feedback for educational practice. Artificial intelligence technology can significantly improve the accuracy of teaching behavior analysis, promote innovation in classroom management and teaching methods, and has broad application prospects.

Key words: Artificial intelligence, Classroom teaching behavior, Behavior analysis, Interactive classroom, Data analysis, Big data