ResearchTasksMultilingual Automatic Speech RecognitionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedCommon Voice Exp. 3 (test)Initial model15.4WER (US)13Feb 26, 2026CommonVoiceQwen2.5-Omni + SICL-AT14.42WER (de)13Feb 26, 2026CommonVoice (test)Proposed (MoE + modified CIF)9.86WER6Jun 11, 2026