Build with accurate medical transcription

Create an API key in the console, export it as OMI_API_KEY, then choose an example. Start with a clip up to 60 seconds so automatic language detection returns inline on every plan. The endpoint is OpenAI-compatible, so existing OpenAI SDK clients only need a different base URL.

curl https://api.omi.health/v1/audio/transcriptions \
  -H "Authorization: Bearer $OMI_API_KEY" \
  -F file=@short-dictation.wav \
  -F model=omi-medical-1 \
  -F language=auto \
  -F response_format=verbose_json

python -m pip install openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.omi.health/v1",
    api_key=os.environ["OMI_API_KEY"],
)

with open("short-dictation.wav", "rb") as audio:
    result = client.audio.transcriptions.create(
        model="omi-medical-1",
        file=audio,
        response_format="verbose_json",
        extra_body={"language": "auto"},
    )

print(result.text)

npm install openai

import fs from "node:fs";
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.omi.health/v1",
  apiKey: process.env.OMI_API_KEY,
});

const result = await client.audio.transcriptions.create({
  model: "omi-medical-1",
  file: fs.createReadStream("short-dictation.wav"),
  response_format: "verbose_json",
  language: "auto",
});

console.log(result.text);

Choose the processing region with the hostname. Every host offers the same API, features, API key, pricing and plan limits. Audio, async jobs and stored results stay in the region that processes the request. Console account records and sign-in are in Frankfurt (EU).

US-only processing: use https://api.us.omi.health and wss://api.us.omi.health/v1/stream to keep every request in the US. Only the hostname changes.

Poll and download async jobs through the poll_url and result URLs the API returns; a job stays in the region that created it. Open realtime connections do not move between regions; reconnect to change host. Choosing api.eu.omi.health or api.us.omi.health is your regional processing instruction under the DPA. Read the routing announcement.

Use verbose_json when you need timestamps, language metadata, model information, or a request ID for support. Use text for plain text and json for the smallest JSON response.

{
  "text": "The patient takes metformin 500 mg twice daily.",
  "language": "en",
  "language_source": "detected",
  "duration": 4.82,
  "model": "omi-medical-1",
  "request_id": "0b0e…",
  "segments": [
    {
      "id": "seg_0001",
      "speaker": "A",
      "text": "The patient takes metformin 500 mg twice daily.",
      "start": 0.22,
      "end": 4.60,
      "confidence": null
    }
  ],
  "metadata": {
    "api_version": "2026-07-01",
    "runtime": "omi-runtime",
    "processing_ms": {"asr": 640, "diarization": 0}
  }
}

• English is the benchmarked, medically evaluated language.
• Spanish, Portuguese, French, German, Dutch, Arabic, and Hindi are available for testing and evaluation, and are not medically evaluated.
• language=auto
• detects the spoken language on short inline requests and long jobs on every plan.
• Encounter vocabulary accepts up to 1,000 terms and selects at most 50 per request; the response separately reports accepted, delivered, corrected, and unresolved terms.
• Speaker labels and word timestamps are added without changing transcript wording.
• Patterns can recover expected dosage and identifier formatting, when enabled on your key, without acting as a source of transcript truth.