ResearchTasksSpeech Naturalness EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSynthesized speech samplesClean3.42Objective MOS5Jun 9, 2026FacestarAudio-Visual Speech Codec78.5Preference (Ours)3Feb 26, 2026