ResearchTasksSpeech-to-SpeechSpeech-to-SpeechFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedURO-Bench Pro TaskGLM-4-Voice55.47Understanding Score12Mar 26, 2026BigBenchAudioStep-Audio R1.1 33B97.6Score3Jul 7, 2026Enterprise Realtime Voice Agent EvaluationCascaded pipeline755TTFA (ms)3Mar 6, 2026