# Voice Cloning on CV3-Eval multilingual voice cloning (test)

Source: [Voice Cloning on CV3-Eval multilingual voice cloning (test)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test>)

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

Data updated: 2026-03-20T08:19:23.059Z

Evaluation of Voice Cloning performance on the test split of the CV3-Eval multilingual dataset, measured using Word Error Rate (WER) across multiple languages (English, Chinese, Japanese, Korean, German, Spanish, French, Italian, Russian) and hard subsets.

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 Cloning |
| Dataset (including any stated split) | CV3-Eval multilingual voice cloning (test) |
| Metric | WER (zh) |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 18 of 18 |
| 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

- [WER (zh)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(zh)>): Lower is better; unit: SCALAR.

- [WER (en)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(en)>): Lower is better; unit: SCALAR.

- [WER (ja)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(ja)>): Lower is better; unit: SCALAR.

- [WER (ko)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(ko)>): Lower is better; unit: SCALAR.

- [WER (de)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(de)>): Lower is better; unit: SCALAR.

- [WER (es)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(es)>): Lower is better; unit: SCALAR.

- [WER (fr)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(fr)>): Lower is better; unit: SCALAR.

- [WER (it)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(it)>): Lower is better; unit: SCALAR.

- [WER (ru)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(ru)>): Lower is better; unit: SCALAR.

- [WER (hard subset, zh)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(hard%20subset%2C%20zh)>): Lower is better; unit: SCALAR.

- [WER (hard subset, en)](<https://www.sota2.com/research/sota/voice-cloning-on-cv3-eval-multilingual-voice-cloning-test.md?metric=WER%20(hard%20subset%2C%20en)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | WER (zh) | Result date | Report paper |
| --- | --- | --- | --- |
| Fish Audio S2 | 2.65 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| CosyVoice3 | 2.89 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| CosyVoice3-0.5B+DiffRO | 2.89 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| Fish Audio S1 | 2.98 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| CosyVoice2 | 3 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| CosyVoice2+DiffRO | 3 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| CosyVoice3 | 3.01 | Not recorded | 69b0f6e92d603e6de3e45b93 |
| CosyVoice3-1.5B+DiffRO | 3.01 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| MOSS-TTS-Local-Transformer | 3.68 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| CosyVoice3-0.5B | 3.89 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| CosyVoice3-1.5B | 3.91 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| MOSS-TTS-Local-Transformer | 3.95 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| CosyVoice2 | 4.08 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| MOSS-TTS | 4.26 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| MOSS-TTS | 4.42 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| Spark-TTS | 5.15 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| F5-TTS | 5.47 | Not recorded | 69bcd48b0636e5bf4afc29f4 |
| GPT-SoVits | 7.34 | Not recorded | 69bcd48b0636e5bf4afc29f4 |

### Fish Audio S2

WER (zh): 2.65. Result ID: 69b1108c6dbb7075b95ad78d.

[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.

### CosyVoice3

WER (zh): 2.89. Result ID: 69b1108c6dbb7075b95ad790.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

Recorded evaluation configuration: {"Model Scale":"0.5B","Refinement":"DiffRO"}. This record does not assert independent reproduction.

### CosyVoice3-0.5B+DiffRO

WER (zh): 2.89. Result ID: 69bd030047f34ce924a6f764.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

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

### Fish Audio S1

WER (zh): 2.98. Result ID: 69b1108c6dbb7075b95ad78e.

[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.

### CosyVoice2

WER (zh): 3. Result ID: 69b1108c6dbb7075b95ad78f.

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

Recorded evaluation configuration: {"Refinement":"DiffRO"}. This record does not assert independent reproduction.

### CosyVoice2+DiffRO

WER (zh): 3. Result ID: 69bd030047f34ce924a6f762.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

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

### CosyVoice3

WER (zh): 3.01. Result ID: 69b1108c6dbb7075b95ad791.

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

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

Recorded evaluation configuration: {"Model Scale":"1.5B","Refinement":"DiffRO"}. This record does not assert independent reproduction.

### CosyVoice3-1.5B+DiffRO

WER (zh): 3.01. Result ID: 69bd030047f34ce924a6f766.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

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

### MOSS-TTS-Local-Transformer

WER (zh): 3.68. Result ID: 69bd030047f34ce924a6f76a.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Implementation](<https://github.com/OpenMOSS/MOSS-TTS>)

Recorded evaluation configuration: {"Mode":"Continuation"}. This record does not assert independent reproduction.

### CosyVoice3-0.5B

WER (zh): 3.89. Result ID: 69bd030047f34ce924a6f763.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

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

### CosyVoice3-1.5B

WER (zh): 3.91. Result ID: 69bd030047f34ce924a6f765.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78dd1e209131cea7ae5b.md>)

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

### MOSS-TTS-Local-Transformer

WER (zh): 3.95. Result ID: 69bd030047f34ce924a6f769.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Implementation](<https://github.com/OpenMOSS/MOSS-TTS>)

Recorded evaluation configuration: {"Mode":"Clone"}. This record does not assert independent reproduction.

### CosyVoice2

WER (zh): 4.08. Result ID: 69bd030047f34ce924a6f761.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

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

### MOSS-TTS

WER (zh): 4.26. Result ID: 69bd030047f34ce924a6f768.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Implementation](<https://github.com/OpenMOSS/MOSS-TTS>)

Recorded evaluation configuration: {"Mode":"Continuation"}. This record does not assert independent reproduction.

### MOSS-TTS

WER (zh): 4.42. Result ID: 69bd030047f34ce924a6f767.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Implementation](<https://github.com/OpenMOSS/MOSS-TTS>)

Recorded evaluation configuration: {"Mode":"Clone"}. This record does not assert independent reproduction.

### Spark-TTS

WER (zh): 5.15. Result ID: 69bd030047f34ce924a6f75f.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77381e209131cea71c24.md>)

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

### F5-TTS

WER (zh): 5.47. Result ID: 69bd030047f34ce924a6f75e.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.md>)

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

### GPT-SoVits

WER (zh): 7.34. Result ID: 69bd030047f34ce924a6f760.

[Reporting paper](<https://www.sota2.com/research/paper/69bcd48b0636e5bf4afc29f4.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.
