Test medical speech on your own audio
Record a clip or upload a file and inspect Omi's transcript before creating an account.
Open the playground →For clinicians who code and teams who ship
Build healthcare voice AI.
Use the medical speech API for the fastest path, or deploy the open model inside your own stack. Start small without choosing a toy tier.
Hosted API
From first request to production.
Build with batch and live medical transcription, speakers, timestamps and vocabulary. Your 25 free hours reset every month; adding a card keeps them and protects continuity.
Open model
Own the speech runtime.
Run omi-medical-edge-1 on Mac, CUDA or CPU under CC-BY-4.0. Inspect it, benchmark it and keep audio entirely inside your environment.
Start here
Useful before your first request.
Short, concrete resources for choosing a model and putting clinical audio into a real product.
Best medical speech-to-text APIs in 2026
Compare the criteria that matter: medical terms, dosages, BAA access, speakers and total price.
Read the guide →HIPAA-compliant speech-to-text checklist
A practical pre-PHI checklist for healthcare builders evaluating a speech API.
Use the checklist →30 systems on the same medical audio
Inspect overall, medical-term, drug-name and dosage accuracy independently.
Explore the benchmark →Omi vs Deepgram for medical speech
Medical benchmark results, feature packaging and pricing mechanics side by side.
Compare Deepgram →Deploy medical ASR locally
Weights, runtime-specific benchmark results and installation instructions for omi-medical-edge-1.
Open the model card →What are you building?
Start with the complete medical API. No card, no reduced-feature sandbox.