# Gesture-to-Speech Synthesis on PATS

Source: [Gesture-to-Speech Synthesis on PATS](<https://www.sota2.com/research/sota/gesture-to-speech-synthesis-on-pats>)

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

Data updated: 2026-03-23T11:28:27.660Z

Evaluation of models performing Gesture-to-Speech Synthesis using the PATS dataset, measured by metrics including Speech Quality and Prosodic Similarity.

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 | Gesture-to-Speech Synthesis |
| Dataset (including any stated split) | PATS |
| Metric | Speech Quality |
| 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

- [Speech Quality](<https://www.sota2.com/research/sota/gesture-to-speech-synthesis-on-pats.md?metric=Speech%20Quality>): Higher is better; unit: PERCENT.

- [Prosodic Similarity](<https://www.sota2.com/research/sota/gesture-to-speech-synthesis-on-pats.md?metric=Prosodic%20Similarity>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Speech Quality | Result date | Report paper |
| --- | --- | --- | --- |
| Gesture2Speech (Multimodal MoE) | 81.48 | Not recorded | 69c0c8e50636e5bf4afc3a0e |
| Gesture2Speech (XTTS v2) | 75.79 | Not recorded | 69c0c8e50636e5bf4afc3a0e |
| Gesture2Speech (H-MoE) | 73.55 | Not recorded | 69c0c8e50636e5bf4afc3a0e |
| Gesture2Speech (Unimodal MoE) | 72.22 | Not recorded | 69c0c8e50636e5bf4afc3a0e |
| Gesture2Speech (GPT-SoVITS) | 70.78 | Not recorded | 69c0c8e50636e5bf4afc3a0e |

### Gesture2Speech (Multimodal MoE)

Speech Quality: 81.48. Result ID: 69c123d247f34ce924a99d6a.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

Recorded evaluation configuration: {"Model Variant":"Multimodal MoE"}. This record does not assert independent reproduction.

### Gesture2Speech (XTTS v2)

Speech Quality: 75.79. Result ID: 69c123d247f34ce924a99d66.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73831e209131cea5cd91.md>)

Recorded evaluation configuration: {"Model Variant":"XTTS v2"}. This record does not assert independent reproduction.

### Gesture2Speech (H-MoE)

Speech Quality: 73.55. Result ID: 69c123d247f34ce924a99d69.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

Recorded evaluation configuration: {"Model Variant":"H-MoE"}. This record does not assert independent reproduction.

### Gesture2Speech (Unimodal MoE)

Speech Quality: 72.22. Result ID: 69c123d247f34ce924a99d68.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

Recorded evaluation configuration: {"Model Variant":"Unimodal MoE"}. This record does not assert independent reproduction.

### Gesture2Speech (GPT-SoVITS)

Speech Quality: 70.78. Result ID: 69c123d247f34ce924a99d67.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e50636e5bf4afc3a0e.md>)

Recorded evaluation configuration: {"Model Variant":"GPT-SoVITS"}. This record does not assert independent reproduction.

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