Fact · Voice and AI
The VoiceAgentBench paper says existing speech benchmarks mostly test isolated skills such as transcription or question answering, and do not systematically test agentic behavior or adversarial robustness.
Checked against its source on October 17, 2026
In the source's own words
“existing speech benchmarks largely focus on isolated capabilities such as transcription or question answering and do not systematically evaluate agentic behavior or adversarial robustness.”
VoiceAgentBench: Are Voice Assistants ready for agentic tasks?, Dhruv Jain et al., arXiv, 2025, Abstract, second sentence.
Open the original- What it measures
- The authors use this gap to motivate a benchmark of spoken queries that require tool calls, multi-step workflows and safety checks.
- What it does not show
- This is the authors' description of the benchmark literature at the time of writing. It is not a survey of every speech benchmark.