# Multi-choice Audio Understanding on MMAU speech

Source: [Multi-choice Audio Understanding on MMAU speech](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-17T07:33:00.399Z

Evaluation of multi-choice audio understanding capabilities using the MMAU speech dataset, reporting specialized metrics like n\_vv, t-acc, and ALAS scores.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Multi-choice Audio Understanding |
| Dataset (including any stated split) | MMAU speech |
| Metric | N\_VV Count |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [N\_VV Count](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=N_VV%20Count>): Higher is better; unit: SCALAR.

- [A-Accuracy](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=A-Accuracy>): Higher is better; unit: PERCENT.

- [T-Accuracy](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=T-Accuracy>): Higher is better; unit: PERCENT.

- [L\* Score](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=L*%20Score>): Higher is better; unit: SCALAR.

- [ALAS-soft (vv)](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=ALAS-soft%20(vv)>): Higher is better; unit: SCALAR.

- [ALAS-soft (all)](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=ALAS-soft%20(all)>): Higher is better; unit: SCALAR.

- [ALAS-hard (vv)](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=ALAS-hard%20(vv)>): Higher is better; unit: PERCENT.

- [ALAS-hard (all)](<https://www.sota2.com/research/sota/multi-choice-audio-understanding-on-mmau-speech.md?metric=ALAS-hard%20(all)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | N\_VV Count | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen-Omni | 201 | Not recorded | 68fa79d91e209131cea7fbe6 |
| AF3 | 176 | Not recorded | 68fa79d91e209131cea7fbe6 |
| Qwen2-Audio | 128 | Not recorded | 68fa79d91e209131cea7fbe6 |
| SALMONN | 102 | Not recorded | 68fa79d91e209131cea7fbe6 |

### Qwen-Omni

N\_VV Count: 201. Result ID: 6a324d93fbc373592c2f809a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79d91e209131cea7fbe6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77a21e209131cea7482c.md>)

Recorded evaluation configuration: {"Backbone":"Streaming-modified Whisper","LLM":"Qwen2.5 Thinker","Adapter":"Learned projection"}. This record does not assert independent reproduction.

### AF3

N\_VV Count: 176. Result ID: 6a324d93fbc373592c2f8098.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79d91e209131cea7fbe6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a4d1e209131cea82504.md>)

Recorded evaluation configuration: {"Backbone":"AF-Whisper","LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction.

### Qwen2-Audio

N\_VV Count: 128. Result ID: 6a324d93fbc373592c2f8099.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79d91e209131cea7fbe6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74281e209131cea6017e.md>)

Recorded evaluation configuration: {"Backbone":"Whisper-large-v3","LLM":"Qwen-7B","Adapter":"None"}. This record does not assert independent reproduction.

### SALMONN

N\_VV Count: 102. Result ID: 6a324d93fbc373592c2f809b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79d91e209131cea7fbe6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d8a.md>)

Recorded evaluation configuration: {"Backbone":"Whisper-v2 + BEATs","LLM":"Vicuna-13B","Adapter":"Q-Former"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.