Advancing Theater and Vocal Research in Chinese Opera for Role-Centric Acoustic and Speech Studies through Fuzzy Topsis Evaluation

To improve the performance assessment and facilitate the protection of Chinese opera, there is a need for a sophisticated knowledge of role-dependent acoustic and speech features. In this paper, a hybrid model combining a Transformer-based deep neural network with a fuzzy multi-criteria decision-making model is proposed for all-around role-based vocal analysis. The Transformer network is used to learn fine-grained phonetic and tonal features between the main opera roles Sheng, Dan, Jing, and Chou, and record slight changes in pitch, tone, and expressiveness. To solve the vagueness and hesitation of subjective judgments on performance characteristics like rhythm, voice quality, and articulation, we apply the Circular T-Spherical Fuzzy Bonferroni Mean (CTSF-BM) operator in a TOPSIS-based decision-making model with essential parameters and notations detailed in Table 1. The integration provided ensures strong linguistic rating aggregation together with interrelation conservation between the performance criteria. Comparative analysis based on traditional fuzzy models proves the superiority in accuracy, flexibility, and interpretability of the proposed technique. In total, the paper brings an intelligent data-driven framework for opera performance evaluation with applicable possibilities in broader fields of music and speech analysis.