497 solo piano pieces comprising synchronized sheet music images, MIDI files, and synthesized audio recordings. The dataset provides precise note-level alignments across visual and auditory modalities for classical music compositions.
Use Cases
- Train cross-modal retrieval models to match audio segments with specific measures in sheet music images
- Develop optical music recognition (OMR) systems using the MIDI ground truth and rendered score images
- Build real-time score-following applications using the temporal mapping between audio playback and visual score positions
Strengths
- 497 solo piano compositions from the classical repertoire
- Note-level synchronization between MIDI events, audio timestamps, and sheet music pixel coordinates
- Includes rendered sheet music images and synthesized audio files for consistent multimodal training