ResearchBenchmarksCross-dataset domain transfer (Visual to Audio) on VGGSound-AVEL 90K AVE to AVVPCross-dataset domain transfer (Visual to Audio) on VGGSound-AVEL 90K AVE to AVVPFollow57.8Segment-level F1 ScoreDCID0.28815.21930.1545.081May 12, 2026Evaluation resultsMethodMethodLinksSegment-level F1 ScoreDCIDPretraining Scale=90KPretraining Scale=90K2026.0557.8MICUPretraining Scale=90KPretraining Scale=90K2026.0554.1CoDAARPretraining Scale=90KPretraining Scale=90K2026.0552.3BaselinePretraining Scale=90KPretraining Scale=90K2026.052.5