# Human Speech Generation on LRS2 (test) Source: [Human Speech Generation on LRS2 (test)]() Collection index: [llms.txt]() Data updated: 2026-05-13T10:44:35.717Z Evaluation of Human Speech Generation performance on the test split of the LRS2 dataset, measured using metrics covering spectral distortion (LSE-D, LSE-C), recognition accuracy (WER), speaker similarity (MCD), speech quality (UTMOS), and generative fidelity (FVD, KVD, CLIP). 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 | Human Speech Generation | | Dataset (including any stated split) | LRS2 (test) | | Metric | LSE-D | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 7 of 7 | | 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](): Lower is better; unit: SCALAR. - [LSE-C](): Lower is better; unit: SCALAR. - [WER](): Lower is better; unit: PERCENT. - [MCD](): Lower is better; unit: SCALAR. - [UTMOS](): Higher is better; unit: SCALAR. - [FVD](): Lower is better; unit: SCALAR. - [KVD](): Lower is better; unit: SCALAR. - [CLIP Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | LSE-D | Result date | Report paper | | --- | --- | --- | --- | | SyncDPO | 7.83 | Not recorded | 6a0405dbdbc836b87d827883 | | SyncDPO | 7.98 | Not recorded | 6a0405dbdbc836b87d827883 | | No Tune | 8.15 | Not recorded | 6a0405dbdbc836b87d827883 | | DPO | 8.19 | Not recorded | 6a0405dbdbc836b87d827883 | | DPO | 8.19 | Not recorded | 6a0405dbdbc836b87d827883 | | SFT | 8.2 | Not recorded | 6a0405dbdbc836b87d827883 | | SFT | 8.39 | Not recorded | 6a0405dbdbc836b87d827883 | ### SyncDPO LSE-D: 7.83. Result ID: 6a04555647f34ce924eaece0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"OOD"}. This record does not assert independent reproduction. ### SyncDPO LSE-D: 7.98. Result ID: 6a04555647f34ce924eaecdd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"ID"}. This record does not assert independent reproduction. ### No Tune LSE-D: 8.15. Result ID: 6a04555647f34ce924eaecda. [Reporting paper]() Recorded evaluation configuration: {"Training Data":"None"}. This record does not assert independent reproduction. ### DPO LSE-D: 8.19. Result ID: 6a04555647f34ce924eaecdc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"ID"}. This record does not assert independent reproduction. ### DPO LSE-D: 8.19. Result ID: 6a04555647f34ce924eaecdf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"OOD"}. This record does not assert independent reproduction. ### SFT LSE-D: 8.2. Result ID: 6a04555647f34ce924eaecde. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"OOD"}. This record does not assert independent reproduction. ### SFT LSE-D: 8.39. Result ID: 6a04555647f34ce924eaecdb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Data":"ID"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.