Measuring voice agents: turn-taking, tool calls and expressive speech
Comments
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Summary
The article benchmarks voice agents on three separate abilities: deciding when a speaker has finished a turn (TurnBench), choosing the correct action from spoken commands (SLURP), and following voice-style instructions (InstructTTSEval).