File size: 3,117 Bytes
539aad9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
---
pretty_name: "ICASSP 2027 Audio Editing Challenge Track 2: Agent Track"
task_categories:
- audio-to-audio
viewer: false
---

<p align="center">
  <a href="https://audio-editing-challenge.github.io/">
    <img src="assets/cover.png" width="900" alt="ICASSP 2027 Audio Editing Challenge">
  </a>
</p>

<h1 align="center">ICASSP 2027 Audio Editing Challenge Track 2: Agent Track</h1>

<p align="center">
  <a href="https://audio-editing-challenge.github.io/"><img src="https://img.shields.io/badge/Challenge-Website-2563EB?style=flat-square" alt="Challenge website"></a>
  <a href="https://github.com/Audio-Editing-Challenge/Audio-Editing-Challenge-Baseline"><img src="https://img.shields.io/badge/Baseline-Code-181717?style=flat-square&logo=github" alt="Baseline code"></a>
  <a href="https://huggingface.co/datasets/BoJack/MMAE"><img src="https://img.shields.io/badge/%F0%9F%A4%97-MMAE%20Dataset-FFD21E?style=flat-square" alt="MMAE dataset on Hugging Face"></a>
</p>

Test inputs for Track 2 of the [ICASSP 2027 Audio Editing Challenge](https://audio-editing-challenge.github.io/). The set has 500 examples. They cover every MMAE audio modality and every complexity category: single, multi-part, multi-instruction, multi-audio, multi-round, and multi-hop.

This release contains only input audio and instructions. The evaluation rubrics stay private until the official results are finalized.

## Files

```text
track2_agent.json     # metadata (id + messages)
wav/<id>/audioN.wav   # input audio
```

```bash
hf download ICASSP2027-AudioEditingChallenge/Track2-Agent --repo-type dataset --local-dir track2
```

## Format

`messages` uses the same chat format as [MMAE](https://github.com/ddlBoJack/MMAE). Each `audio_url` is relative to the dataset root.

```json
{
  "id": "<sample_id>",
  "messages": [
    {"role": "user", "content": [
      {"type": "text", "text": "<editing instruction>"},
      {"type": "audio", "audio_url": "wav/<sample_id>/audio1.wav"}
    ]}
  ]
}
```

- Multi-audio examples list several `audio` entries in one turn.
- Multi-round examples have several user turns. Apply them in order and submit only the final audio.

## Submission

Each JSON entry is one example. Submit exactly one edited 16 kHz mono WAV per `id` (500 in total), plus a `submission.jsonl` with one line per example:

```json
{"id": "<sample_id>", "audio_path": "audio/<sample_id>.wav"}
```

Package `submission.jsonl` together with the `audio/` folder and upload it to the [leaderboard](https://audio-editing-challenge.github.io/leaderboard/). The full rules are on the [Track 2](https://audio-editing-challenge.github.io/track2/) page.

## Baseline

The [baseline](https://github.com/Audio-Editing-Challenge/Audio-Editing-Challenge-Baseline) uses DeepSeek-V4-Flash as a router over DSP tools, SAM-Audio-Large, and AuK with the Prompt Enhancer. It was scored on this test set with the Qwen3-Omni-30B-A3B-Instruct judge:

| IFR | CR | EMR |
| :---: | :---: | :---: |
| 44.13% | 76.09% | 8.00% |

Teams are ranked by EMR. Ties are broken by IFR, then by CR.

## Contact

zhikangniu@sjtu.edu.cn, tuwenming@sjtu.edu.cn