# Speech-driven talking face generation on VoxCeleb (test)

Source: [Speech-driven talking face generation on VoxCeleb (test)](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test>)

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

Data updated: 2026-03-04T12:14:08.295Z

Evaluation of speech-driven talking face generation models on the test split of the VoxCeleb dataset, assessed using metrics including LSE-D, LSE-C, LPIPS, SSIM, PSNR, FID, and MAD.

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-driven talking face generation |
| Dataset (including any stated split) | VoxCeleb (test) |
| Metric | LSE-D |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 8 of 8 |
| 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

- [LSE-D](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=LSE-D>): Lower is better; unit: SCALAR.

- [LSE-C](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=LSE-C>): Lower is better; unit: SCALAR.

- [LPIPS](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=LPIPS>): Lower is better; unit: SCALAR.

- [SSIM](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=SSIM>): Higher is better; unit: SCALAR.

- [PSNR](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=PSNR>): Higher is better; unit: SCALAR.

- [FID](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=FID>): Lower is better; unit: SCALAR.

- [MAD](<https://www.sota2.com/research/sota/speech-driven-talking-face-generation-on-voxceleb-test.md?metric=MAD>): Lower is better; unit: SCALAR.

## Recorded results

| Method | LSE-D | Result date | Report paper |
| --- | --- | --- | --- |
| HRTF | 6.61 | Not recorded | 690836f03ef7ccd74e5e5345 |
| HRTF-SCFP | 7.08 | Not recorded | 690836f03ef7ccd74e5e5345 |
| VideoRetalking | 9.44 | Not recorded | 690836f03ef7ccd74e5e5345 |
| Aniportait | 9.63 | Not recorded | 690836f03ef7ccd74e5e5345 |
| SyncTalk | 11.46 | Not recorded | 690836f03ef7ccd74e5e5345 |
| Hallo | 12.43 | Not recorded | 690836f03ef7ccd74e5e5345 |
| Anitalker | 14.76 | Not recorded | 690836f03ef7ccd74e5e5345 |
| Real3D-portrait | 16.6 | Not recorded | 690836f03ef7ccd74e5e5345 |

### HRTF

LSE-D: 6.61. Result ID: 69a821fc6dbb7075b94e33ef.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

Recorded evaluation configuration: {"Venue":"-"}. This record does not assert independent reproduction.

### HRTF-SCFP

LSE-D: 7.08. Result ID: 69a821fc6dbb7075b94e33f0.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

Recorded evaluation configuration: {"Venue":"-"}. This record does not assert independent reproduction.

### VideoRetalking

LSE-D: 9.44. Result ID: 69a821fc6dbb7075b94e33ee.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e0e1e209131cea3dbd2.md>)

Recorded evaluation configuration: {"Venue":"SIGGRAPH 2022"}. This record does not assert independent reproduction.

### Aniportait

LSE-D: 9.63. Result ID: 69a821fc6dbb7075b94e33e9.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72b31e209131cea5805f.md>)

Recorded evaluation configuration: {"Venue":"ECCV 2024"}. This record does not assert independent reproduction.

### SyncTalk

LSE-D: 11.46. Result ID: 69a821fc6dbb7075b94e33eb.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730f1e209131cea5a370.md>)

[Implementation](<https://ziqiaopeng.github.io/synctalk>)

Recorded evaluation configuration: {"Venue":"CVPR 2024"}. This record does not assert independent reproduction.

### Hallo

LSE-D: 12.43. Result ID: 69a821fc6dbb7075b94e33ec.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b31e209131cea5dfcf.md>)

Recorded evaluation configuration: {"Venue":"arXiv 2024.7"}. This record does not assert independent reproduction.

### Anitalker

LSE-D: 14.76. Result ID: 69a821fc6dbb7075b94e33ed.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73141e209131cea5a5b7.md>)

Recorded evaluation configuration: {"Venue":"ACM MM 2024"}. This record does not assert independent reproduction.

### Real3D-portrait

LSE-D: 16.6. Result ID: 69a821fc6dbb7075b94e33ea.

[Reporting paper](<https://www.sota2.com/research/paper/690836f03ef7ccd74e5e5345.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72a31e209131cea57ba3.md>)

Recorded evaluation configuration: {"Venue":"ICLR 2024"}. This record does not assert independent reproduction.

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