The CAL500 Expansion (CAL500exp) dataset is an enriched version of the CAL500 music information retrieval dataset. CAL500exp is designed to facilitate music auto-tagging in a smaller temporal scale. The dataset consists of the same songs split into 3,223 acoustically homogenous segments of 3 to 16 seconds. The tag labels are annotated in the segment level instead of track level. The annotations were obtained from annotators with strong music background.
Source: Towards time-varying music auto-tagging based on CAL500 expansionPaper | Code | Results | Date | Stars |
---|