Research on music interactive teaching mode based on wireless sensor network and multimodal data fusion

Abstract To address the issues of insufficient personalized feedback and poor real-time performance in traditional music teaching, this paper proposes a music interactive teaching mode based on wireless sensor networks and multimodal data fusion. By deploying wireless sensor nodes such as motion sensors, acoustic sensors, and biological sensors, multimodal data (including students’ performance actions, audio, and physiological states) during teaching are collected in real-time. The Gaussian-Bayesian algorithm is used for fusion to obtain multimodal fused data with abnormal interference removed, which is then input into the cloud music teaching resource sharing platform. A weighted matrix factorization algorithm is applied to quantify students’ preferences for erroneous performance segments, generating personalized error-correction content push results. These results are interactively distributed to designated students via the cloud platform, enabling music interactive teaching. Experimental results show that after multimodal data fusion, the transmission data volume of wireless sensor nodes is reduced by 78.8 %, and the sensor energy consumption is lowered to 0.33 J, ensuring long-term stable operation. The average hit rate of personalized push reaches 95.2 %, accurately matching students’ preferences for erroneous performance segments. The high frequency of information interaction in teaching forms a closed-loop of “collection-fusion-push-feedback” interactive teaching, providing an efficient and precise technical solution for the digital interactive transformation of music teaching.