ResearchBenchmarksAudio-driven Talking Face Generation on User StudyAudio-driven Talking Face Generation on User StudyFollow4.6Lip-sync AccuracyPC-Talk1.1682.0592.953.841Mar 18, 2025Evaluation resultsMethodMethodLinksLip-sync AccuracyImage QualityTemporal ConsistencyEmotional ExpressivenessOverall RealismPC-Talk2025.034.64.84.64.94.8Hallo-v22025.033.843.4—3.8Wav2Lip2025.033.31.91.9—1.9ED-Talk2025.033.11.42.42.31.6MuseTalk2025.032.22.63—3.4Echomimic2025.031.93.43.5—3EAT2025.031.32.92.62.62.1