call recording transcription webhook

Here’s a simple, practical call recording transcription webhook pattern you can use.

What it does

When a call recording is ready, your telephony provider sends a webhook to your server. Your server then:

1. receives the recording URL / ID,
2. downloads the audio,
3. sends it to a transcription service,
4. stores or returns the transcript.

---

Example webhook flow

1) Provider sends webhook

Typical payload might look like:

{
  "event": "call.recording.completed",
  "call_id": "abc123",
  "recording_id": "rec_789",
  "recording_url": "https://provider.com/recordings/rec_789.mp3",
  "duration": 245,
  "created_at": "2026-10-14T12:34:56Z"
}

---

2) Example webhook endpoint

Node.js / Express

import express from "express";
import bodyParser from "body-parser";

const app = express();
app.use(bodyParser.json());

app.post("/webhooks/call-recording", async (req, res) => {
  try {
    const { event, recording_url, recording_id, call_id } = req.body;

    if (event !== "call.recording.completed") {
      return res.status(200).json({ ok: true, ignored: true });
    }

    // TODO: verify signature from provider
    // TODO: download recording_url
    // TODO: send audio to transcription API
    // TODO: save transcript in DB

    console.log("Processing recording:", { call_id, recording_id, recording_url });

    res.status(200).json({ ok: true });
  } catch (err) {
    console.error(err);
    res.status(500).json({ ok: false, error: "Internal error" });
  }
});

app.listen(3000, () => console.log("Server listening on 3000"));

---

3) Python / FastAPI example

from fastapi import FastAPI, Request

app = FastAPI()

@app.post("/webhooks/call-recording")
async def call_recording_webhook(request: Request):
    data = await request.json()

    if data.get("event") != "call.recording.completed":
        return {"ok": True, "ignored": True}

    recording_url = data.get("recording_url")
    recording_id = data.get("recording_id")
    call_id = data.get("call_id")

    # TODO: verify signature
    # TODO: download recording
    # TODO: transcribe audio
    # TODO: save transcript

    return {"ok": True}

---

Important production considerations

• Verify webhook signatures to prevent spoofing.
• Respond quickly (usually within a few seconds). Offload transcription to a background job queue.
• Make processing idempotent so duplicate webhook deliveries don’t create duplicate transcripts.
• Handle retries gracefully.
• Secure audio URLs if they require auth.
• Store raw event payloads for debugging/auditing.

---

If you want a transcription step

You can send the downloaded audio to services like:

• OpenAI Speech-to-Text
• Deepgram
• AssemblyAI
• AWS Transcribe
• Google Speech-to-Text

---

Example background job concept

1. Webhook receives event
2. Save job to queue (e.g. BullMQ, Celery, SQS)
3. Worker downloads audio
4. Worker transcribes it
5. Worker saves transcript
6. Optionally notify your app/customer

---

If you want, I can generate a ready-to-use webhook endpoint for a specific stack like:

• Node.js + Express
• Python + FastAPI
• PHP / Laravel
• Next.js
• AWS Lambda
• Twilio / Vonage / Plivo / Zoom Phone