Free Music Archive Audio Features and Metadata

FMA is a dataset for music analysis, containing audio tracks and associated metadata. It includes features for music genre classification, audio analysis, and music information retrieval. The dataset was created by researchers and is hosted on the UCI Machine Learning Repository.

Use Cases

• Predict music genre labels from audio features like spectral centroid and MFCCs.
• Analyze track metadata such as artist, album, and play counts for music recommendation systems.
• Train models for audio similarity search using precomputed features.
• Perform music genre classification using the provided hierarchical genre taxonomy.

Strengths

• Includes a hierarchical genre taxonomy for multi-level classification.
• Provides precomputed audio features for analysis and modeling.

Limitations

• Unknown total number of audio tracks and rows.
• Potential class imbalance in genre distribution not specified.

Provenance

Source

UCI Machine Learning Repository.

Collection Method

Aggregated from the Free Music Archive, with audio features and metadata extracted.