Table 5 Evaluation measures for music segmentation with respect to different boundaries. F-measure (F), precision (P), and recall (R) are estimated for all boundaries (A) and boundaries coming with instrument (I), key (K), or tempo (T) change. Abbreviations for the feature groups: MelS: the Mel spectrum; SemS: the semitone spectrum; Fluct: fluctuation patterns. Best values are marked with bold font
From: AAM: a dataset of Artificial Audio Multitracks for diverse music information retrieval tasks
Meas. | MFCCs | MelS | SemS | Fluct | CNN |
|---|---|---|---|---|---|
\(F_{\textrm{A}}\) | 0.646 | 0.577 | 0.615 | 0.470 | 0.914 |
\(P_{\textrm{A}}\) | 0.568 | 0.473 | 0.506 | 0.397 | 0.950 |
\(R_{\textrm{A}}\) | 0.810 | 0.803 | 0.850 | 0.615 | 0.899 |
\(F_{\textrm{I}}\) | 0.600 | 0.531 | 0.571 | 0.418 | 0.925 |
\(P_{\textrm{I}}\) | 0.484 | 0.401 | 0.432 | 0.322 | 0.947 |
\(R_{\textrm{I}}\) | 0.871 | 0.870 | 0.926 | 0.644 | 0.921 |
\(F_{\textrm{K}}\) | 0.520 | 0.469 | 0.510 | 0.374 | 0.822 |
\(P_{\textrm{K}}\) | 0.399 | 0.338 | 0.370 | 0.274 | 0.798 |
\(R_{\textrm{K}}\) | 0.863 | 0.887 | 0.951 | 0.666 | 0.901 |
\(F_{\textrm{T}}\) | 0.387 | 0.357 | 0.377 | 0.299 | 0.846 |
\(P_{\textrm{T}}\) | 0.272 | 0.237 | 0.251 | 0.201 | 0.895 |
\(R_{\textrm{T}}\) | 0.840 | 0.912 | 0.947 | 0.718 | 0.857 |