GTZAN Musical Genre Audio Dataset with 10 Genres

The GTZAN dataset contains 1,000 audio tracks for musical genre classification, each 30 seconds long. It includes 10 distinct genres, with 100 tracks per genre, all formatted as 22,050Hz Mono 16-bit WAV files.

Use Cases

• Train a genre classifier using the 10 genre labels on 1,000 audio tracks.
• Analyze audio signal features across the 10 genres, such as blues, classical, and rock.
• Benchmark music information retrieval models on the 30-second audio track format.

Strengths

• Contains 1,000 audio tracks, providing a foundational corpus for genre classification.
• Includes 10 distinct musical genres, each represented by 100 tracks for balanced representation.
• All audio tracks are consistently formatted as 30-second, 22,050Hz Mono 16-bit WAV files.

Limitations

• The dataset is relatively small with only 1,000 tracks, which may limit model generalization.
• The 30-second track length may not capture full song structures or longer musical patterns.
• Genre labels are broad and may not reflect sub-genres or modern musical styles.

Provenance

Source

marsyas

Collection Method

Dataset for musical genre classification of audio signals.

Freshness

Last updated on 2023-11-26.