# Voice Interaction on VoiceBench

Source: [Voice Interaction on VoiceBench](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench>)

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

Data updated: 2026-04-29T07:11:39.761Z

Evaluation of model performance on the VoiceBench dataset, focusing on general Voice Interaction capabilities, measured across a comprehensive suite of LLM, QA, and specialized voice evaluation metrics.

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 | Voice Interaction |
| Dataset (including any stated split) | VoiceBench |
| Metric | VoiceBench Average Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

- [VoiceBench Average Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=VoiceBench%20Average%20Score>): Higher is better; unit: PERCENT.

- [Overall Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

- [IFEval Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=IFEval%20Score>): Higher is better; unit: PERCENT.

- [BBH Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=BBH%20Score>): Higher is better; unit: PERCENT.

- [Adversarial Benchmark Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=Adversarial%20Benchmark%20Score>): Higher is better; unit: PERCENT.

- [AlpacaEval Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=AlpacaEval%20Score>): Higher is better; unit: PERCENT.

- [CommonEval Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=CommonEval%20Score>): Higher is better; unit: PERCENT.

- [WildVoice Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=WildVoice%20Score>): Higher is better; unit: PERCENT.

- [OpenBookQA Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=OpenBookQA%20Score>): Higher is better; unit: PERCENT.

- [MMSU Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=MMSU%20Score>): Higher is better; unit: PERCENT.

- [SD-QA Score](<https://www.sota2.com/research/sota/voice-interaction-on-voicebench.md?metric=SD-QA%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | VoiceBench Average Score | Result date | Report paper |
| --- | --- | --- | --- |
| Nemotron 3 Nano Omni | 89.4 | Not recorded | 69f1907355a3c32af89fb98b |
| Qwen3-Omni | 88.8 | Not recorded | 69f1907355a3c32af89fb98b |
| Qwen3.5-Omni | 87.8 | Not recorded | 69f1907355a3c32af89fb98b |

### Nemotron 3 Nano Omni

VoiceBench Average Score: 89.4. Result ID: 69f1ae5847f34ce924d7ae85.

[Reporting paper](<https://www.sota2.com/research/paper/69f1907355a3c32af89fb98b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f1907355a3c32af89fb98b.md>)

Recorded evaluation configuration: {"Open-Source":"true","Size":"30B-A3B","Reasoning mode":"Reasoning on"}. This record does not assert independent reproduction.

### Qwen3-Omni

VoiceBench Average Score: 88.8. Result ID: 69f1ae5847f34ce924d7ae86.

[Reporting paper](<https://www.sota2.com/research/paper/69f1907355a3c32af89fb98b.md>)

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

Recorded evaluation configuration: {"Open-Source":"true","Size":"30B-A3B","Reasoning mode":"Reasoning on"}. This record does not assert independent reproduction.

### Qwen3.5-Omni

VoiceBench Average Score: 87.8. Result ID: 69f1ae5847f34ce924d7ae87.

[Reporting paper](<https://www.sota2.com/research/paper/69f1907355a3c32af89fb98b.md>)

Recorded evaluation configuration: {"Open-Source":"false","Size":"Flash","Reasoning mode":"Reasoning on"}. This record does not assert independent reproduction.

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