AssemblyAI launches STT-3.7-Realtime-Agent for voice agents, 13.2% entity error beats 26 rivals
harnoor-assemblyai · reddit · 2026-11-19
AssemblyAI released STT-3.7-Realtime-Agent, a realtime speech-to-text model for voice agents supporting 32 languages. It targets failure modes that cause agents to take wrong actions: background voices, dropped short replies, misheard numbers, and premature turn endings.
On the company's benchmark of 12,460 scripted caller clips, it scored 13.2% entity error, beating the next-best of 26 tested systems, Soniox Realtime v6 (15.2%). On long-form English, Meta's model posted lower WER: 7.37% vs 8.35%. Benchmarks are self-run, with testing between August and November.
More from coding & agent
- Ultrafast mode review: great for math proofs, not recommended for coding — gnukeith · 2026-11-19
- Speed up CI for AI-written code: caching, faster runners, test sharding — rseroter · 2026-11-19
- Agentcard raises $8M to let agents buy from 5,000+ stores via Stripe Link — jeff_weinstein · 2026-11-19
- hstack open-sources one-command self-hosting for the Hermes agent — TheMoonMidas · 2026-11-19
- 16 AI models ran produce buying for 90 days — only 3 beat a simple reorder formula — ycombinator · 2026-11-19
- DeepLearning.AI Launches Short Course on Tabular Foundation Models as Agent Tools — allenainie · 2026-11-19