# Speaker Similarity on MiniMax-MLS (test)

Source: [Speaker Similarity on MiniMax-MLS (test)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test>)

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

Data updated: 2026-06-08T07:04:44.445Z

Evaluation of speaker similarity performance on the test split of the MiniMax-MLS dataset, reporting scores across 24 different languages.

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 | Speaker similarity |
| Dataset (including any stated split) | MiniMax-MLS (test) |
| Metric | Speaker Similarity Score (Chinese) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Speaker Similarity Score (Chinese)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Chinese)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Arabic)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Arabic)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Cantonese)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Cantonese)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Czech)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Czech)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Dutch)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Dutch)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (English)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(English)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Finnish)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Finnish)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (French)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(French)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (German)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(German)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Greek)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Greek)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Hindi)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Hindi)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Indonesian)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Indonesian)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Italian)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Italian)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Japanese)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Japanese)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Korean)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Korean)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Polish)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Polish)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Portuguese)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Portuguese)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Romanian)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Romanian)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Russian)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Russian)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Spanish)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Spanish)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Thai)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Thai)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Turkish)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Turkish)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Ukrainian)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Ukrainian)>): Higher is better; unit: PERCENT.

- [Speaker Similarity Score (Vietnamese)](<https://www.sota2.com/research/sota/speaker-similarity-on-minimax-mls-test.md?metric=Speaker%20Similarity%20Score%20(Vietnamese)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Speaker Similarity Score (Chinese) | Result date | Report paper |
| --- | --- | --- | --- |
| VoxCPM2 | 82.5 | Not recorded | 6a264c5590966ade919be7d1 |
| Fish Audio S2 | 81.6 | Not recorded | 6a264c5590966ade919be7d1 |
| Qwen3-TTS | 79.9 | Not recorded | 6a264c5590966ade919be7d1 |
| Minimax | 78 | Not recorded | 6a264c5590966ade919be7d1 |
| ElevenLabs | 67.7 | Not recorded | 6a264c5590966ade919be7d1 |

### VoxCPM2

Speaker Similarity Score (Chinese): 82.5. Result ID: 6a266989fbc373592c22a98e.

[Reporting paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Fish Audio S2

Speaker Similarity Score (Chinese): 81.6. Result ID: 6a266989fbc373592c22a98d.

[Reporting paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b0f6e92d603e6de3e45b93.md>)

[Implementation](<https://github.com/fishaudio/fish-speech>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-TTS

Speaker Similarity Score (Chinese): 79.9. Result ID: 6a266989fbc373592c22a98c.

[Reporting paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6973006ab2be0ed85a78c8c5.md>)

[Implementation](<https://github.com/QwenLM/Qwen3-TTS>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Minimax

Speaker Similarity Score (Chinese): 78. Result ID: 6a266989fbc373592c22a98a.

[Reporting paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78891e209131cea78eba.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ElevenLabs

Speaker Similarity Score (Chinese): 67.7. Result ID: 6a266989fbc373592c22a98b.

[Reporting paper](<https://www.sota2.com/research/paper/6a264c5590966ade919be7d1.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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