Name: Nova-3 Medical API
Brand: Deepgram

Overview

Deepgram Nova-3 Medical is a domain-specialized variant of Nova-3, fine-tuned on clinical and healthcare audio datasets. It delivers significantly higher accuracy for medical terminology — prescription names, diagnoses, procedures, anatomical terms — making it the optimal choice for clinical documentation, telehealth platforms, EHR integrations, and medical dictation tools.

Technical Specifications

Performance Benchmarks

Significantly lower word error rate on medical terminology vs. general-purpose models.
Accurately recognizes drug names, ICD-10 diagnoses, and procedural terms.
Real-time streaming with sub-300ms latency for live clinical workflows.
Handles noisy clinical environments: background equipment sounds, phone audio.
Designed to reduce manual review burden in medical documentation.

Architecture Breakdown

Nova-3 Medical is built on the same end-to-end deep learning engine as Nova-3, with an additional fine-tuning stage on curated medical audio corpora. The model's vocabulary and language model are biased toward clinical terminology, enabling reliable transcription of complex medical speech without custom vocabulary configuration.

Pricing

$0.00559 / min

Core Features & Capabilities

Medical Terminology Recognition: Accurate transcription of drug names, diagnoses, symptoms, and procedures.
Streaming Transcription: Real-time clinical dictation via WebSocket.
Speaker Diarization: Labels clinician and patient voices separately.
Smart Formatting: Handles medical number formats, dosages, and dates.
Entity Detection: Extracts medical entities from audio for downstream processing.
Custom Vocabulary (keyterm): Add institution-specific or rare drug names to further boost accuracy.
Dictation Mode: Optimized for voice-driven clinical note-taking workflows.
Filler Word Detection: Removes hesitation markers common in dictation speech.

Comparison with Other Models

VS Deepgram Nova-3: Nova-3 Medical is fine-tuned for clinical audio and priced lower per minute; Nova-3 is the general-purpose English model for non-medical use cases.
VS Deepgram Nova-3 General: Nova-3 Medical is domain-specialized for healthcare; Nova-3 General covers 30+ languages for general multilingual workflows.
VS AssemblyAI Slam-1: Nova-3 Medical focuses on clinical ASR accuracy; Slam-1 provides semantic understanding and prompt-based customization for enterprise transcription workflows.

‍

Use cases

Clinical documentation: Nova-3 Medical is documented for real-time clinical note-taking and streamlining doctor-patient conversations into structured notes ready for EHR integration.

Prescription workflows: It is documented for capturing prescription instructions by voice, converting unstructured dictation into structured medication orders, and routing them for pharmacy handling.

Appointment scheduling and patient registration: It is documented for voice-driven scheduling, registration, cancellations, and follow-ups in healthcare settings.

Telehealth and clinical decision support: It is described as suited to real-time healthcare applications such as telemedicine, digital health platforms, and clinical decision support systems.

How to use Nova-3 Medical API

Install any OpenAI-compatible SDK, point it at api.aimlapi.com/v1, and set the model to deepgram/nova-3-medical.

import requests, time

headers = {"Authorization": "Bearer " + AIMLAPI_KEY}
job = requests.post(
    "https://api.aimlapi.com/v1/stt/create",
    headers=headers,
    json={
      "model": "deepgram/nova-3-medical",
      "url": "https://example.com/audio.mp3"
    },
).json()
gid = job["generation_id"]

while True:
    res = requests.get(f"https://api.aimlapi.com/v1/stt/{gid}", headers=headers).json()
    if res.get("status") in ("completed", "error", "failed"):
        break
    time.sleep(3)
print(res)

const headers = {
  Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
  "Content-Type": "application/json",
};
const job = await (await fetch("https://api.aimlapi.com/v1/stt/create", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "model": "deepgram/nova-3-medical",
    "url": "https://example.com/audio.mp3"
  }),
})).json();

let res;
do {
  await new Promise((r) => setTimeout(r, 3000));
  res = await (await fetch(`https://api.aimlapi.com/v1/stt/${job.generation_id}`, { headers })).json();
} while (!["completed", "error", "failed"].includes(res.status));
console.log(res);

# submit the job — the response contains "generation_id"
curl -X POST https://api.aimlapi.com/v1/stt/create \
  -H "Authorization: Bearer $AIMLAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepgram/nova-3-medical","url":"https://example.com/audio.mp3"}'

# then poll for the result until it is ready
curl "https://api.aimlapi.com/v1/stt/{generation_id}" -H "Authorization: Bearer $AIMLAPI_KEY"

OpenAI-compatible — swap the base URL and it works with your existing SDK.