Explore high-quality datasets for your AI and machine learning projects.
MusicCaps is a dataset of 5,521 music excerpts, each paired with an English aspect list and a free‑text caption written by musicians. Captions focus on acoustic characteristics rather than metadata such as artist name. The dataset is released as a CSV file containing YouTube video IDs and start/end timestamps; users must download the corresponding YouTube videos and clip them according to the timestamps.
