ElevenLabs launches HIPAA-eligible Scribe v2 Medical speech-to-text model
Summary
ElevenLabs launches the HIPAA-eligible Scribe v2 Medical speech-to-text model on ElevenAPI, offering enterprise customers Zero Retention Mode and posting a 4.9% word error rate across English, French, and German on Corti’s MedDictate benchmark.
Key Points
- ElevenLabs releases Scribe v2 Medical, a clinical-audio Speech to Text model now generally available on ElevenAPI under the model ID scribe_v2_medical.
- On Corti’s MedDictate benchmark, Scribe v2 Medical records a 4.9% overall word error rate across English, French, and German, about 35% lower than base Scribe v2’s 7.6%.
- The model is HIPAA-eligible for enterprise customers with Business Associate Agreements and Zero Retention Mode, which deletes audio input and text output immediately after API requests finish.