Skip to main content
This guide takes a recorded consultation and returns a structured clinical note. The whole flow is five HTTP calls — no SDK required.

​
Prerequisites

1

Get API credentials

Sign up on Eka, then in the Developer Console go to Manage API Credentials and create a client. Save the client_id and client_secret — the secret is shown only once.
2

Have an audio file ready

Any consultation or dictation recording in webm, wav, mp3, m4a, mp4 or ogg.

​
The complete flow

BASE=https://api.eka.care

# 1. Authenticate — exchange credentials for an access token
TOKEN=$(curl -s -X POST $BASE/connect-auth/v1/account/login \
  -H "Content-Type: application/json" \
  -d '{"client_id":"YOUR_CLIENT_ID","client_secret":"YOUR_CLIENT_SECRET"}' \
  | jq -r .access_token)

# 2. Create a session
SESSION_ID=$(curl -s -X POST $BASE/voice/v1/sessions \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{
    "language_hint": ["en"],
    "model": "pro",
    "templates": ["clinical_notes_template"],
    "upload_type": "single"
  }' | jq -r .session_id)

# 3. Upload the audio file (content type is detected from the extension)
curl -s -X POST "$BASE/voice/v1/sessions/$SESSION_ID/audio/audio_0.webm" \
  -H "Authorization: Bearer $TOKEN" \
  --data-binary @consultation.webm

# 4. End the session — processing starts
curl -s -X POST "$BASE/voice/v1/sessions/$SESSION_ID/end" \
  -H "Authorization: Bearer $TOKEN"

# 5. Poll until done (HTTP 202 = still processing, 200 = completed)
curl -s "$BASE/voice/v1/sessions/$SESSION_ID" \
  -H "Authorization: Bearer $TOKEN"

import requests, time

BASE = "https://api.eka.care"

# 1. Authenticate — exchange credentials for an access token
token = requests.post(f"{BASE}/connect-auth/v1/account/login", json={
    "client_id": "YOUR_CLIENT_ID",
    "client_secret": "YOUR_CLIENT_SECRET",
}).json()["access_token"]
auth = {"Authorization": f"Bearer {token}"}

# 2. Create a session
session = requests.post(f"{BASE}/voice/v1/sessions", headers=auth, json={
    "language_hint": ["en"],
    "model": "pro",
    "templates": ["clinical_notes_template"],
    "upload_type": "single",
}).json()
sid = session["session_id"]

# 3. Upload the audio file (content type is detected from the extension)
with open("consultation.webm", "rb") as f:
    requests.post(
        f"{BASE}/voice/v1/sessions/{sid}/audio/audio_0.webm",
        headers=auth,
        data=f,
    )

# 4. End the session — processing starts
requests.post(f"{BASE}/voice/v1/sessions/{sid}/end", headers=auth)

# 5. Poll until done (HTTP 202 = still processing, 200 = completed)
while True:
    r = requests.get(f"{BASE}/voice/v1/sessions/{sid}", headers=auth)
    if r.status_code != 202:
        break
    time.sleep(1)

result = r.json()
print(result["transcript"])
print(result["templates"])   # structured note per requested template

import { readFile } from "node:fs/promises";

const BASE = "https://api.eka.care";

// 1. Authenticate — exchange credentials for an access token
const { access_token } = await fetch(`${BASE}/connect-auth/v1/account/login`, {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    client_id: "YOUR_CLIENT_ID",
    client_secret: "YOUR_CLIENT_SECRET",
  }),
}).then((r) => r.json());
const auth = { Authorization: `Bearer ${access_token}` };

// 2. Create a session
const session = await fetch(`${BASE}/voice/v1/sessions`, {
  method: "POST",
  headers: { ...auth, "Content-Type": "application/json" },
  body: JSON.stringify({
    language_hint: ["en"],
    model: "pro",
    templates: ["clinical_notes_template"],
    upload_type: "single",
  }),
}).then((r) => r.json());

// 3. Upload the audio file (content type is detected from the extension)
await fetch(
  `${BASE}/voice/v1/sessions/${session.session_id}/audio/audio_0.webm`,
  {
    method: "POST",
    headers: auth,
    body: await readFile("consultation.webm"),
  }
);

// 4. End the session — processing starts
await fetch(`${BASE}/voice/v1/sessions/${session.session_id}/end`, {
  method: "POST",
  headers: auth,
});

// 5. Poll until done (HTTP 202 = still processing, 200 = completed)
let result;
while (true) {
  const r = await fetch(`${BASE}/voice/v1/sessions/${session.session_id}`, {
    headers: auth,
  });
  if (r.status !== 202) { result = await r.json(); break; }
  await new Promise((s) => setTimeout(s, 1000));
}

console.log(result.transcript);
console.log(result.templates); // structured note per requested template
The completed response contains the transcript and one structured document per requested template: 
{
  "session_id": "ses_abc123def456",
  "status": "completed",
  "transcript": "Patient presents with cough and fever...",
  "templates": [
    {
      "clinical_notes_template": {
        "status": "success",
        "document_id": "doc_abc123",
        "data": { "chief_complaint": "cough and fever", "assessment": "Acute bronchitis" }
      }
    }
  ]
}
upload_type: "single" sends one complete audio file — max 10 MB via the API. For larger recordings, create the session with upload_type: "chunked" and upload sequential chunks (audio_0.webm, audio_1.webm, …), or use an SDK which records, chunks and uploads automatically. For live scribing, use upload_type: "stream" with the WebSocket API.

​
Next steps

Try it in the browser

Paste your credentials and call every API from the docs — no code needed.

Create your own template

Define your note format in markdown and pass its ID in templates.

Explore the Sessions API

Discovery, streaming, session updates, webhooks and status codes.

Get notified via webhook

Skip polling — receive a callback when the note is ready.

​
Prefer an SDK?

SDKs handle microphone capture, voice-activity detection, chunking, retries and polling for you: 
import { getEkaScribeInstance } from "@eka-care/ekascribe-ts-sdk";

const ekascribe = getEkaScribeInstance({ access_token: "YOUR_ACCESS_TOKEN" });

await ekascribe.initTransaction({
  mode: "consultation",
  input_language: ["en-IN"],
  output_format_template: [{ template_id: "clinical_notes_template" }],
  txn_id: "unique-session-id",
  transfer: "vaded",
  model_type: "pro",
});
await ekascribe.startRecording();
// ... consultation happens ...
await ekascribe.endRecording();

const result = await ekascribe.pollSessionOutput({ txn_id: "unique-session-id" });
Available for TypeScript, Python, Java, Android and iOS.