Research on Role Orientation and Situation Construction of Contextualized Music Performance in the Background of Artificial Intelligence

Abstract

Abstract In this paper, the Skip-gram model is used to process the main melody information of contextualized music, combining feature extraction and Schenkel analysis to extract the main melody note sets and vectors. By applying the short-time Fourier transform (STFT) to the audio signal, the spectral center of mass and irregularities can be calculated to represent the emotional features of the music. To obtain music features based on wavelet coefficients, the music signal is decomposed, and each scale’s signal features are counted individually. Finally, by taking the musical elements as independent variables and the perceived results of pleasure and activation in the contextual model as dependent variables, we compared the variability of the dynamic associations between emotional perceptions and the musical elements among different contextualized music pieces. The results showed that the overall differences in emotional cognition across musical performances ranged between ±0.5, with the mean short-term energy amplitude of relatively cheerful music ranging between ±0.2, which was smaller than that of angry (±0.3) and sad (±0.5) music. In this study, musical works were accurately characterized, and their expressive and infectious power was effectively enhanced through contextual construction.