Research on Optimization of Stage Presentation Effects and Intelligent Assisted Decision-Making for Music Performances Based on Multimodal Data Fusion

The integration of digital technologies into performing arts has created opportunities for quantitative evaluation and intelligent optimization of stage presentation effects. This study proposes a multimodal-data-fusion-based framework for stage-effect evaluation and intelligent decision support in music performance scenarios. A multimodal acquisition platform is developed to synchronously collect electromyography signals, inertial measurement unit data, audio spectral information, spatial trajectories, and audience physiological responses. To address temporal asynchrony and heterogeneous feature distributions, an attention-based feature-fusion strategy is employed to construct a multidimensional stage-effect evaluation model. Furthermore, a reinforcement-learning-based decision-support mechanism dynamically coordinates lighting, acoustic, and visual presentation elements. Experimental results demonstrate that the multimodal framework significantly improves effect-prediction accuracy and enhances audience immersion and emotional engagement. The proposed approach provides methodological references for multimodal signal fusion, intelligent sensing, human-centered interaction, and adaptive communication environments.