---
license: cc-by-nc-sa-4.0
language:
- en
pretty_name: STEMMA-Bench
size_categories:
- n<1K
task_categories:
- audio-text-to-text
- question-answering
tags:
- music
- multi-audio
- benchmark
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test.jsonl
dataset_info:
  features:
  - name: id
    dtype: string
  - name: task
    dtype: string
  - name: question
    dtype: string
  - name: answer
    dtype: string
  - name: audios
    sequence: string
  splits:
  - name: test
    num_examples: 300
---

# STEMMA-Bench

<p align="center">
  <a href="https://arxiv.org/abs/2610.11884">📄 Paper</a> &nbsp;|&nbsp;
  <a href="https://github.com/hoyso48/STEMMA">💻 Code</a> &nbsp;|&nbsp;
  <a href="https://huggingface.co/datasets/hoyso48/STEMMA-Instruct">🤗 STEMMA-Instruct</a> &nbsp;|&nbsp;
  <a href="https://huggingface.co/hoyso48/STEMMA-MF2601">🤗 STEMMA-MF2601</a> &nbsp;|&nbsp;
  <a href="https://huggingface.co/hoyso48/STEMMA-MOSS-Music">🤗 STEMMA-MOSS-Music</a>
</p>

Multi-audio music relation benchmark from *STEMMA: Song-to-Stem Multi-Audio Reasoning for Large Audio Language
Models*. Each item has an anchor (`Audio A`) and four candidates (`Candidate B`–`E`), one correct (chance 25%). Clips
are sections and native stems of multitrack recordings; every item was checked by listening.

| | |
|---|---|
| Items | 300 (100 each from MUSDB18-HQ, MoisesDB, MedleyDB) |
| Structural / attribute | 127 / 173 |
| Tasks | 14 |

## Format

| Field | Description |
|---|---|
| `id` | item ID |
| `task` | `<relation>_anchor_candidates_mcq` |
| `question` | prompt with five `<audio>` (Audio A, Candidate B–E) |
| `answer` | correct letter (`B`–`E`) |
| `audios` | audio IDs, in prompt order |

Structural tasks: `same_track`, `same_track_block`, `adjacent_master`, `adjacent_block`, `same_section_cross_role`,
`parent_child`; the rest are attribute tasks (key, tempo, instrument, genre). `_block` tasks use stems. Accuracy is
reported per group; predicting `A` counts as wrong.

## Audio

Audio is not included. Audio IDs are `<corpus>/<track>/<section>[/<stem>]`; `metadata/audio.jsonl` gives each clip's
source files in the original corpus, start/end time, and whether summed stems are peak-normalized. 210 MedleyDB section
mixes come from the mixture files in
[seungheondoh/cmd-medleydb-metadata](https://huggingface.co/datasets/seungheondoh/cmd-medleydb-metadata)
(revision `99e4ae2`).

## Citation

```bibtex
@misc{sohn2026stemma,
  title         = {{STEMMA}: Song-to-Stem Multi-Audio Reasoning for Large Audio Language Models},
  author        = {Sohn, Hoyeol and Kim, Wonil and Kim, Keunhyoung and Kum, Sangeun and Kim, Taehyoung and Moon, Dongjoo and Charoenchob, Theerasak and Weerapang, Teeratep and Lee, Jongpil and Nam, Juhan},
  year          = {2026},
  eprint        = {2610.11884},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SD},
  url           = {https://arxiv.org/abs/2610.11884}
}
```

## License

CC BY-NC-SA 4.0. The audio remains subject to the MUSDB18-HQ, MoisesDB, and MedleyDB terms.