Classification of Ethnic Music Styles Combining Timbre Clustering and Rhythmic Pattern Matching

Current research on ethnic music style classification faces challenges related to timbre overlap in multi-instrument recordings and unstable representation of rhythmic structures, which reduce the reliability of style discrimination. This paper proposes a classification framework combining timbre clustering with rhythmic pattern matching. Robust spectral–temporal feature representation and sequential pattern analysis are increasingly important in intelligent acoustic sensing and electromagnetic signal processing applications, providing methodological support for complex signal interpretation across engineering domains. The proposed framework constructs independent yet complementary representations for timbre distribution and rhythm organization, enabling spectral and temporal ambiguities to be jointly constrained within a unified classification process. Spectral features are extracted using short-time Fourier transform (STFT), followed by the computation of 13-dimensional MFCCs, first-order differential coefficients, spectral centroids, roll-off rates, and spectral flux features. After principal component analysis, Gaussian mixture models are employed for timbre clustering to isolate instrument-related acoustic components. Rhythm features are subsequently generated from timbre-cluster-dependent energy envelopes and aligned with ethnic rhythm templates through Dynamic Time Warping, allowing robust similarity measurement under nonlinear temporal variations. The timbre cluster probability distribution and rhythm matching descriptors are fused into a multidimensional feature vector and classified using a support vector machine. Experimental results demonstrate that the proposed model achieves an overall accuracy of 90.9% and an average F1 score of 91.3% for six ethnic music styles, validating the effectiveness and stability of jointly modeling timbre clustering and rhythmic pattern matching. The proposed framework provides a reliable strategy for complex spectral–temporal signal analysis and offers potential methodological insights for advanced electromagnetic sensing and intelligent signal interpretation.