# Meetings end, transcripts appear: RealtimeKit post-meeting transcription is now GA!

Hey there, it's Shiichan! Today I found a nice little update around meetings, so let me tell you about it.

## What was announced?

On the Cloudflare Changelog, [RealtimeKit](https://developers.cloudflare.com/realtime/realtimekit/) announced that its post-meeting transcription is now Generally Available (GA). Once a meeting wraps up, it automatically produces transcript files and AI summaries for you. The transcription itself is handled by [Whisper Large v3 Turbo](https://developers.cloudflare.com/workers-ai/models/whisper-large-v3-turbo/) running on [Workers AI](https://developers.cloudflare.com/workers-ai/).

## The story so far

Turning meeting recordings into transcripts used to mean standing up your own transcription infrastructure or wiring in an external service. RealtimeKit is a platform for embedding WebRTC video and audio calls, and now that managed transcription and summaries ride on top of it, all that plumbing simply goes away.

## What changes

When you create a meeting, you just set `transcribe_on_end: true` and `summarize_on_end: true`, and the transcript and summary run automatically after the meeting ends. You can receive the results via [webhook](https://developers.cloudflare.com/realtime/realtimekit/webhooks/) (the `meeting.transcript` and `meeting.summary` events) or fetch them later through the [REST API](https://developers.cloudflare.com/realtime/realtimekit/ai/summary/#rest-api). Writing meeting notes and following up on tasks gets a whole lot easier.

## Dive Deep

Transcript [output formats](https://developers.cloudflare.com/realtime/realtimekit/ai/transcription/#output-formats) come in four flavors: CSV, JSON, SRT, and VTT. The language is auto-detected, with support for Whisper language codes.

For summaries, you can fine-tune parameters like word-count limits, Markdown formatting, and meeting-type classification. You can even regenerate just the summary later through the API.

If you want transcription in real time, there's a separate real-time option powered by [Deepgram Nova-3](https://developers.cloudflare.com/workers-ai/models/nova-3/). Being able to pick based on your use case is lovely. If you want the full details, check out the [transcription docs](https://developers.cloudflare.com/realtime/realtimekit/ai/transcription/).

## Wrap-up

- RealtimeKit post-meeting transcription is now GA
- Whisper Large v3 Turbo on Workers AI auto-generates transcripts plus AI summaries
- Outputs are CSV, JSON, SRT, and VTT; grab results via webhook or REST API
- Just flip on `transcribe_on_end` / `summarize_on_end` when creating a meeting

This one is perfect for anyone who wants to automate notes and summaries without running their own transcription infrastructure!