Music Examples with English Aspect Lists and Captions

The MusicCaps dataset contains 5,521 music examples. Each example is labeled with an English aspect list and a free-text caption written by musicians.

Use Cases

• Train a text-to-music generation model using the English aspect list as a structured prompt.
• Develop a music captioning model by learning to generate the free-text caption from audio features.
• Analyze the relationship between structured aspect lists and descriptive free-text captions for music understanding.
• Fine-tune a language model for music description using the 5,521 musician-written captions.

Strengths

• Contains 5,521 music examples.
• Each example includes both a structured aspect list and a free-text caption.
• Labels are created by musicians, providing expert annotations.

Limitations

• Dataset size of 5,521 examples may be limited for training large generative models.
• Limited to English language descriptions, which may not generalize to other languages.
• The specific audio features or file formats are not detailed in the provided input.

Provenance

Source

google

Collection Method

Each music example is labeled with an aspect list and caption written by musicians.

Freshness

Last updated on 2023-03-08.