| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Time-controlled instruction-following | InstructS2S long (30s–1min) | MAE5.16 | 8 | |
| Time-controlled instruction-following | InstructS2S short (10s–30s) | MAE3.16 | 8 | |
| Speech-to-speech instruction-following | InstructS2S Eval | S2S Acc3.47 | 4 | |
| Interruption | InstructS2S-Eval | Overlap (s)1.1 | 3 | |
| Turn-taking | InstructS2S Eval | SR@3s92 | 3 | |
| English-to-Chinese Speech-to-Speech Conversation | InstructS2S Eval | GPT Score (T)4.29 | 2 |