A METHOD FOR ANALYZING, CLASSIFYING, AND MATCHING MUSICAL WORKS ACCORDING TO THEIR ACOUSTIC CONTENT.

Abstract

The invention enables the computer-aided analysis of the acoustic content of musical works, the statistical processing of the obtained acoustic features, the preprocessing of the dataset to improve the accuracy of the analysis, the implementation of dimensionality reduction processes to ensure more efficient representation of high-dimensional acoustic data, the determination of acoustic similarities between works using multidimensional similarity analyses, the grouping of musical works into natural acoustic clusters based on the determined similarities without requiring any pre-labeling or user guidance, the validation of the created cluster structures using various statistical evaluation criteria, the association of the validated clusters with appropriate emotional categories by considering their common acoustic characteristics, and the determination of the distinctive acoustic features contributing to the formation of each cluster.This encompasses an integrated music analysis, grouping, and emotional matching method that allows for the predictable reporting of analysis results, the verification of method reliability by comparing computer-assisted analysis results with evaluations derived from human perception, and the identification of congruences or dissonances between lyrics and acoustic structure by jointly evaluating emotional assessments derived from the lyrical content of musical works with the results obtained from acoustic analyses. This enables not only the evaluation of musical works under specific classes but also the revelation of natural acoustic relationships, the acquisition of explainable and verifiable analysis results, and adaptability to different music genres, datasets, and computer systems.