PERiScoPe: Aligned Piano Scores and Performances for Music AI

PERiScoPe is a large-scale dataset of aligned piano scores and performances developed for the SyMuPe research project. It combines and processes open-source collections like (n)ASAP and ATEPP with curated web-collected MIDI performances. The dataset was created by SyMuPe and was last updated on the platform in January 2026.

Use Cases

• Training models for expressive music performance generation based on aligned score-performance pairs.
• Researching the relationship between musical scores and their expressive interpretations.
• Developing tools for automatic music transcription and score alignment.
• Analyzing performance styles and nuances in piano music.

Strengths

• Dataset is described as 'large-scale', indicating substantial size.
• Integrates multiple established open-source collections ((n)ASAP, ATEPP) with curated web data.
• Data has been aligned and cleaned to produce parallel score and performance note sequences.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown, limiting suitability assessment.
• Data may reflect source bias inherent to the combined open-source and web-collected MIDI performances.

Provenance

Source

Combines open-source collections ((n)ASAP, ATEPP) and curated web-collected MIDI performances.

Collection Method

Collected, aligned, and cleaned to produce parallel score and performance note sequences.

Freshness

Last updated 2026-01-10 13:54:09; freshness should be verified.