Real-Time Feedback System for Stage Performance Using Multi-Modal Deep Learning Techniques

In recent years, with the rapid development of intelligent educational technology, the application of deep learning in the field of art teaching has received widespread attention. Stage performance teaching, especially bel canto education, has long faced the difficulties of strong teaching subjectivity, delayed feedback and lack of scientific evaluation methods. To solve these problems, this paper proposes a real-time feedback system based on multimodal deep learning, which combines convolutional neural network (CNN) and long short-term memory network (LSTM) to achieve comprehensive analysis and instant feedback of students' voice and movement performance. feedback. The system collects audio signals and motion data, uses multimodal fusion technology to build a feedback model, and combines the attention mechanism to optimize the weight distribution of sound and motion features. The experimental results show that the students in the experimental group are significantly better than those in the control group in many indicators such as pitch stability, timbre consistency, movement fluency and stability. Among them, pitch stability and timbre consistency increased by 18% and 22% respectively, and movement fluency and stability scores increased by more than 40%. The overall performance score of the experimental group increased by 30%, while the control group increased by less than 10%. These results verify the effectiveness of the multimodal deep learning feedback system and its application value in teaching practice. The main contribution of this paper is to introduce multimodal deep learning technology into Bel Canto teaching and propose a new paradigm of real-time feedback.