# Speech Synthesis on Minimax Multilingual (test)

Source: [Speech Synthesis on Minimax Multilingual (test)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test>)

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

Data updated: 2026-06-08T07:04:44.582Z

Evaluation of Speech Synthesis performance on the test split of the Minimax Multilingual dataset, measured across numerous languages using Word Error Rate (WER) and Speech Intelligibility Metric (SIM).

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 | Speech Synthesis |
| Dataset (including any stated split) | Minimax Multilingual (test) |
| Metric | Chinese WER (%) |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 9 of 9 |
| 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

- [Chinese WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Chinese%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Arabic WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Arabic%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Arabic SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Arabic%20SIM>): Higher is better; unit: SCALAR.

- [Cantonese WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Cantonese%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Cantonese SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Cantonese%20SIM>): Higher is better; unit: SCALAR.

- [Chinese SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Chinese%20SIM>): Higher is better; unit: SCALAR.

- [Czech WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Czech%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Czech SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Czech%20SIM>): Higher is better; unit: SCALAR.

- [Dutch WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Dutch%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Dutch SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Dutch%20SIM>): Higher is better; unit: SCALAR.

- [English WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=English%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [English SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=English%20SIM>): Higher is better; unit: SCALAR.

- [Finnish WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Finnish%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Finnish SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Finnish%20SIM>): Higher is better; unit: SCALAR.

- [French WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=French%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [French SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=French%20SIM>): Higher is better; unit: SCALAR.

- [German WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=German%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [German SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=German%20SIM>): Higher is better; unit: SCALAR.

- [Greek WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Greek%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Greek SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Greek%20SIM>): Higher is better; unit: SCALAR.

- [Hindi WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Hindi%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Hindi SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Hindi%20SIM>): Higher is better; unit: SCALAR.

- [Indonesian WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Indonesian%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Indonesian SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Indonesian%20SIM>): Higher is better; unit: SCALAR.

- [Italian WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Italian%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Italian SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Italian%20SIM>): Higher is better; unit: SCALAR.

- [Japanese WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Japanese%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Japanese SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Japanese%20SIM>): Higher is better; unit: SCALAR.

- [Korean WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Korean%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Korean SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Korean%20SIM>): Higher is better; unit: SCALAR.

- [Polish WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Polish%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Polish SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Polish%20SIM>): Higher is better; unit: SCALAR.

- [Portuguese WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Portuguese%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Portuguese SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Portuguese%20SIM>): Higher is better; unit: SCALAR.

- [Romanian WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Romanian%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Romanian SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Romanian%20SIM>): Higher is better; unit: SCALAR.

- [Russian WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Russian%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Russian SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Russian%20SIM>): Higher is better; unit: SCALAR.

- [Spanish WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Spanish%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Spanish SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Spanish%20SIM>): Higher is better; unit: SCALAR.

- [Thai WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Thai%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Thai SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Thai%20SIM>): Higher is better; unit: SCALAR.

- [Turkish WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Turkish%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Turkish SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Turkish%20SIM>): Higher is better; unit: SCALAR.

- [Ukrainian WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Ukrainian%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Ukrainian SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Ukrainian%20SIM>): Higher is better; unit: SCALAR.

- [Vietnamese WER (%)](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Vietnamese%20WER%20(%25)>): Lower is better; unit: PERCENT.

- [Vietnamese SIM](<https://www.sota2.com/research/sota/speech-synthesis-on-minimax-multilingual-test.md?metric=Vietnamese%20SIM>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Chinese WER (%) | Result date | Report paper |
| --- | --- | --- | --- |
| Fish Audio S2 | 0.73 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| Fish Audio S2 | 0.73 | Not recorded | 6a264c5590966ade919be7d1 |
| Qwen3-TTS | 0.93 | Not recorded | 6a264c5590966ade919be7d1 |
| Fish Audio S1 | 0.98 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| VoxCPM2 | 1.14 | Not recorded | 6a264c5590966ade919be7d1 |
| Minimax | 2.25 | Not recorded | 6a264c5590966ade919be7d1 |
| MiniMax | 2.252 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| ElevenLabs | 16.026 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| ElevenLabs | 16.03 | Not recorded | 6a264c5590966ade919be7d1 |

### Fish Audio S2

Chinese WER (%): 0.73. Result ID: 69b1108c6dbb7075b95ad77b.

[Reporting paper](<https://www.sota2.com/research/paper/69b0f6e92d603e6de3e45b93.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.

### Fish Audio S2

Chinese WER (%): 0.73. Result ID: 6a266989fbc373592c22a988.

[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: {"ASR evaluation model":"Whisper-large-v3"}. This record does not assert independent reproduction.

### Qwen3-TTS

Chinese WER (%): 0.93. Result ID: 6a266989fbc373592c22a987.

[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: {"ASR evaluation model":"Whisper-large-v3"}. This record does not assert independent reproduction.

### Fish Audio S1

Chinese WER (%): 0.98. Result ID: 69b1108c6dbb7075b95ad77c.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa75991e209131cea68dc0.md>)

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

### VoxCPM2

Chinese WER (%): 1.14. Result ID: 6a266989fbc373592c22a989.

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

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

Recorded evaluation configuration: {"ASR evaluation model":"Whisper-large-v3"}. This record does not assert independent reproduction.

### Minimax

Chinese WER (%): 2.25. Result ID: 6a266989fbc373592c22a985.

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

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

Recorded evaluation configuration: {"ASR evaluation model":"Whisper-large-v3"}. This record does not assert independent reproduction.

### MiniMax

Chinese WER (%): 2.252. Result ID: 69b1108c6dbb7075b95ad779.

[Reporting paper](<https://www.sota2.com/research/paper/69b0f6e92d603e6de3e45b93.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

Chinese WER (%): 16.026. Result ID: 69b1108c6dbb7075b95ad77a.

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

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

### ElevenLabs

Chinese WER (%): 16.03. Result ID: 6a266989fbc373592c22a986.

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

Recorded evaluation configuration: {"ASR evaluation model":"Whisper-large-v3"}. This record does not assert independent reproduction.

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