Published July 31, 2026 | Version v1
Dataset Open

H2H Music Improv: Human-to-Human Music Improvisation Dataset

Authors/Creators

  • 1. Georgia Institute of Technology
  • 2. ROR icon Massachusetts Institute of Technology

Description

We present the H2H (Human-to-Human) Music Improvisation Dataset, containing six hours of audio-visual expert duo improvisations with clean per-player stems and per-player annotations, in which each performer annotates both their own musical navigation intentions and their peception of their partner’s.

The work has been published in the Proceedings of the Society for Music Information Retrieval Conference (ISMIR) 2026. Project page could be found at https://h2himprov.github.io/. 

Files

annotation.csv

Files (93.5 GB)

Name Size
md5:fac608398a5890ce108bcc48624bb73f
120.3 kB Preview Download
md5:3e74b982c445344df0fa8d3f379fcf5f
18.6 GB Preview Download
md5:80da51cea1c53d20fa6f96f13d73d2a7
37.8 GB Preview Download
md5:1b4579b363ed4f2a0c3f0537087079ce
37.1 GB Preview Download

Additional details

Software

Repository URL
https://h2himprov.github.io/