Vibeleaderboard
← All Intel
Intel / article

Scribe V3 Live Medical

Source
elevenlabs.io
Author
ElevenLabs editorial sitemap
Date
Why it matters

Voice agents in healthcare can now use a real-time transcription model tuned for drug names and clinical terms, with up to 10,000 custom terms, reducing misheard medications during live calls.

Key takeaways · AI-distilled
  • ElevenLabs built Scribe v3 Live Medical for live calls after seeing a general speech model transcribe the diabetes drug sitagliptin as the antidepressant citalopram, the kind of mishearing that puts the wrong drug on a refill request.
  • In ElevenLabs' internal tests on PriMock57, 57 unscripted mock primary-care consultations over noisy phone audio, the model had a 12.8% word error rate, 17% fewer errors than the next medical model it evaluated.
  • Teams can load up to 10,000 custom terms, such as clinic names or a drug portfolio, which ElevenLabs says cut medical-term errors by up to 48% in its tests.
  • The model waits when callers read IDs or birth dates in pieces instead of replying mid-number, and follows mid-call language switches across more than 90 languages; ElevenLabs says it matches Scribe v3 Live's roughly 150 ms latency.
  • It is HIPAA-eligible only for enterprise customers with a business associate agreement and Zero Retention Mode enabled, and every cited comes from ElevenLabs' own testing of selected providers.
Terms in this piece · Glossary
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
  • benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
Read the source elevenlabs.io
More from ElevenLabs editorial sitemap
Recommended reads
Comments

Checking sign-in…

Loading comments…