# General Audio Understanding on MMSU 1.0 (test)

Source: [General Audio Understanding on MMSU 1.0 (test)](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test>)

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

Data updated: 2026-02-26T05:57:47.202Z

Evaluation of general audio understanding capabilities, covering perception and reasoning across semantic, phonological, and paralinguistic dimensions, using the test split of MMSU version 1.0.

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 | General Audio Understanding |
| Dataset (including any stated split) | MMSU 1.0 (test) |
| Metric | Perception Semantics |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [Perception Semantics](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Perception%20Semantics>): Higher is better; unit: PERCENT.

- [Perception Phonology](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Perception%20Phonology>): Higher is better; unit: PERCENT.

- [Perception Paralinguistics](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Perception%20Paralinguistics>): Higher is better; unit: PERCENT.

- [Perception Average](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Perception%20Average>): Higher is better; unit: PERCENT.

- [Reasoning Semantics](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Reasoning%20Semantics>): Higher is better; unit: PERCENT.

- [Reasoning Phonology](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Reasoning%20Phonology>): Higher is better; unit: PERCENT.

- [Reasoning Paralinguistics](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Reasoning%20Paralinguistics>): Higher is better; unit: PERCENT.

- [Reasoning Average](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Reasoning%20Average>): Higher is better; unit: PERCENT.

- [Overall Accuracy](<https://www.sota2.com/research/sota/general-audio-understanding-on-mmsu-1-0-test.md?metric=Overall%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Perception Semantics | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen3-Omni | 72.13 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Qwen2.5-Omni | 61.11 | Not recorded | 69802fd9a8b01e76cc8e181f |
| DIFFA-2 | 60.63 | Not recorded | 69802fd9a8b01e76cc8e181f |
| GPT-4o-Audio | 59.7 | Not recorded | 69802fd9a8b01e76cc8e181f |
| DIFFA-2 | 59.69 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Kimi-Audio | 57.64 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Qwen-Audio-Chat | 57.21 | Not recorded | 69802fd9a8b01e76cc8e181f |
| MiniCPM-O | 56.56 | Not recorded | 69802fd9a8b01e76cc8e181f |
| DIFFA | 52.67 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Qwen2-Audio | 52.14 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Gemini 2.0 Flash | 47.17 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Baichuan-Audio | 39.63 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Phi-4-multimodal | 38.72 | Not recorded | 69802fd9a8b01e76cc8e181f |
| Salmonn | 31.55 | Not recorded | 69802fd9a8b01e76cc8e181f |
| GLM-4-Voice | 27.8 | Not recorded | 69802fd9a8b01e76cc8e181f |
| LTU | 21.34 | Not recorded | 69802fd9a8b01e76cc8e181f |

### Qwen3-Omni

Perception Semantics: 72.13. Result ID: 699a004e5698665ec48250c7.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b881e209131cea88f3d.md>)

Recorded evaluation configuration: {"Size":"30B-A3B"}. This record does not assert independent reproduction.

### Qwen2.5-Omni

Perception Semantics: 61.11. Result ID: 699a004e5698665ec48250cd.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

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

Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction.

### DIFFA-2

Perception Semantics: 60.63. Result ID: 699a004e5698665ec48250ca.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### GPT-4o-Audio

Perception Semantics: 59.7. Result ID: 699a004e5698665ec48250c8.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Size":null}. This record does not assert independent reproduction.

### DIFFA-2

Perception Semantics: 59.69. Result ID: 699a004e5698665ec48250cb.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

Recorded evaluation configuration: {"Size":"8B","Decoding Strategy":"factor-based parallel decoding"}. This record does not assert independent reproduction.

### Kimi-Audio

Perception Semantics: 57.64. Result ID: 699a004e5698665ec48250cc.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78431e209131cea77803.md>)

Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction.

### Qwen-Audio-Chat

Perception Semantics: 57.21. Result ID: 699a004e5698665ec48250d1.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71861e209131cea51c5b.md>)

[Implementation](<https://github.com/QwenLM/Qwen-Audio>)

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### MiniCPM-O

Perception Semantics: 56.56. Result ID: 699a004e5698665ec48250ce.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### DIFFA

Perception Semantics: 52.67. Result ID: 699a004e5698665ec48250cf.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad11e209131cea854c5.md>)

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### Qwen2-Audio

Perception Semantics: 52.14. Result ID: 699a004e5698665ec48250d0.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

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

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### Gemini 2.0 Flash

Perception Semantics: 47.17. Result ID: 699a004e5698665ec48250c9.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78731e209131cea78622.md>)

Recorded evaluation configuration: {"Size":null}. This record does not assert independent reproduction.

### Baichuan-Audio

Perception Semantics: 39.63. Result ID: 699a004e5698665ec48250d3.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771a1e209131cea71007.md>)

Recorded evaluation configuration: {"Size":"11B"}. This record does not assert independent reproduction.

### Phi-4-multimodal

Perception Semantics: 38.72. Result ID: 699a004e5698665ec48250d2.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction.

### Salmonn

Perception Semantics: 31.55. Result ID: 699a004e5698665ec48250d5.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

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

Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction.

### GLM-4-Voice

Perception Semantics: 27.8. Result ID: 699a004e5698665ec48250d4.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75f91e209131cea6abba.md>)

Recorded evaluation configuration: {"Size":"9B"}. This record does not assert independent reproduction.

### LTU

Perception Semantics: 21.34. Result ID: 699a004e5698665ec48250d6.

[Reporting paper](<https://www.sota2.com/research/paper/69802fd9a8b01e76cc8e181f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72431e209131cea555aa.md>)

[Implementation](<https://github.com/YuanGongND/ltu>)

Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction.

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