ResearchTasksUtterance-level classificationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedARCH (RAVDESS, EMOVO, AudioMNIST, SLURP) post-VQ 57DTM-Codec@50Hz37.85RAVDESS Accuracy15Jun 30, 2026