# Reference-image-conditioned joint audio-video generation on User Study (test) Source: [Reference-image-conditioned joint audio-video generation on User Study (test)]() Collection index: [llms.txt]() Data updated: 2026-04-23T10:04:17.687Z Human evaluation of models performing reference-image-conditioned joint audio-video generation on the test split of the User Study dataset, assessed using metrics covering lip-sync accuracy, facial expression realism, action naturalness, text alignment, subject alignment, and visual quality. 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 | Reference-image-conditioned joint audio-video generation | | Dataset (including any stated split) | User Study (test) | | Metric | Lip-Sync Accuracy | | Unit | SCALAR | | 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 - [Lip-Sync Accuracy](): Higher is better; unit: SCALAR. - [Facial Expression Realism](): Higher is better; unit: SCALAR. - [Action Naturalness](): Higher is better; unit: SCALAR. - [Text Alignment](): Higher is better; unit: SCALAR. - [Subject Alignment](): Higher is better; unit: SCALAR. - [Visual Quality](): Higher is better; unit: SCALAR. - [Overall Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | Lip-Sync Accuracy | Result date | Report paper | | --- | --- | --- | --- | | MMControl | 3.41 | Not recorded | 69e85bff7d0172140ec81326 | | Hallo3 | 3.07 | Not recorded | 69e85bff7d0172140ec81326 | | HunyuanCustom | 2.9 | Not recorded | 69e85bff7d0172140ec81326 | | SadTalker | 2.35 | Not recorded | 69e85bff7d0172140ec81326 | | AniPortrait | 1.53 | Not recorded | 69e85bff7d0172140ec81326 | ### MMControl Lip-Sync Accuracy: 3.41. Result ID: 69e9ee8e47f34ce924d1a242. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"Reference image only"}. This record does not assert independent reproduction. ### Hallo3 Lip-Sync Accuracy: 3.07. Result ID: 69e9ee8e47f34ce924d1a23e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"Ground-truth audio and first video frame"}. This record does not assert independent reproduction. ### HunyuanCustom Lip-Sync Accuracy: 2.9. Result ID: 69e9ee8e47f34ce924d1a23f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"Ground-truth audio and first video frame"}. This record does not assert independent reproduction. ### SadTalker Lip-Sync Accuracy: 2.35. Result ID: 69e9ee8e47f34ce924d1a240. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"Ground-truth audio and first video frame"}. This record does not assert independent reproduction. ### AniPortrait Lip-Sync Accuracy: 1.53. Result ID: 69e9ee8e47f34ce924d1a241. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"Ground-truth audio and first video frame"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.