Connect your app to Cyron
Cyron speaks the LiveKit protocol. Use the official LiveKit SDKs with your Cyron server URL and API key — nothing else changes.
1Get your credentials
Create a project in the dashboard. Each project has its own API keys and an isolated set of rooms.
| Variable | Value |
|---|---|
| LIVEKIT_URL | wss://live.cyron.app |
| LIVEKIT_API_KEY | Dashboard → API keys |
| LIVEKIT_API_SECRET | Dashboard → API keys |
2Mint access tokens on your server
Clients join rooms with short-lived JWTs signed by your API secret. Generate them on your backend with any LiveKit server SDK.
import { AccessToken } from 'livekit-server-sdk';const at = new AccessToken(process.env.LIVEKIT_API_KEY, process.env.LIVEKIT_API_SECRET, {identity: 'user-42',name: 'Ayşe',});at.addGrant({ roomJoin: true, room: 'design-review' });const token = await at.toJwt();
3Connect from the client
Point the LiveKit client SDK at your Cyron server URL. Tracks, data messages and reconnection work exactly as with LiveKit.
import { Room, RoomEvent } from 'livekit-client';const room = new Room({ adaptiveStream: true, dynacast: true });room.on(RoomEvent.TrackSubscribed, (track) => document.body.append(track.attach()));await room.connect('wss://live.cyron.app', token);await room.localParticipant.enableCameraAndMicrophone();
4Manage rooms from your backend
The RoomService API (create, list, delete rooms, manage participants) is available at the same host over HTTPS.
import { RoomServiceClient } from 'livekit-server-sdk';const rooms = new RoomServiceClient('https://live.cyron.app', apiKey, apiSecret);await rooms.createRoom({ name: 'design-review', emptyTimeout: 300 });const list = await rooms.listRooms();await rooms.removeParticipant('design-review', 'user-42');
5Receive webhooks
Set a webhook URL in project settings. Events are signed with your API key and verify with LiveKit's WebhookReceiver.
import { WebhookReceiver } from 'livekit-server-sdk';const receiver = new WebhookReceiver(apiKey, apiSecret);app.post('/livekit/webhook', express.raw({ type: 'application/webhook+json' }), async (req, res) => {const event = await receiver.receive(req.body.toString(), req.get('Authorization'));console.log(event.event, event.room?.name); // room_started, participant_joined, ...res.sendStatus(200);});
Run agents
Build agents with the LiveKit Agents framework (Python or Node), run them on your own servers and point them at Cyron with your project's credentials. Cyron sends them jobs for your project's rooms only, and the dashboard's Agents page lists the connected workers and their sessions.
LIVEKIT_URL=wss://live.cyron.appLIVEKIT_API_KEY=<your API key>LIVEKIT_API_SECRET=<your API secret># python agent.py start · node agent.js start
A worker without an agent name joins every new room of your project. With agent_name set, it joins only where you dispatch it: with the AgentDispatch API, or with roomConfig.agents in a participant's token. Agents are participants, so their minutes count towards your plan.
import { AgentDispatchClient } from 'livekit-server-sdk';const dispatch = new AgentDispatchClient('https://live.cyron.app', apiKey, apiSecret);// an agent registered with agent_name: 'support-bot'await dispatch.createDispatch('design-review', 'support-bot', { metadata: '{"lang":"tr"}' });
LiveKit Cloud's own services (hosted agents, LiveKit Inference, enhanced noise cancellation) are not available: use your own speech-to-text, LLM and text-to-speech providers.
Live captions and translation
On Scale and Enterprise, every participant can read what is said, live, in the language they choose: Cyron transcribes each speaker and translates them into each reader's language. Speech already in a reader's language is only transcribed. 60 languages are available, among them English, Turkish, Spanish, Hindi, Arabic and Chinese.
A participant turns captions on by setting the participant attribute cyron.captions to a language code (ISO 639-1, e.g. tr), and off with an empty string: in their token's attributes, from the client with setAttributes (the token needs canUpdateOwnMetadata), or from your server with RoomService.UpdateParticipant. A hidden Cyron participant then joins the room within a few seconds; it doesn't appear in participant lists and isn't counted as a participant.
// server: let participants set their own attributesat.addGrant({ roomJoin: true, room: 'design-review', canUpdateOwnMetadata: true });// client: captions in Turkish ('' turns them off)await room.localParticipant.setAttributes({ 'cyron.captions': 'tr' });room.on(RoomEvent.DataReceived, (payload, _participant, _kind, topic) => {if (topic !== 'cyron.captions') return;const msg = JSON.parse(new TextDecoder().decode(payload));// { type: 'caption', id, speaker, language, text, final }if (msg.type === 'caption') showCaption(msg.id, msg.speaker, msg.text, msg.final);// { type: 'status', language, state: 'active' | 'limit' | 'unsupported' | 'quota' | 'unavailable' }if (msg.type === 'status' && msg.state !== 'active') showNotice(msg.state);// { type: 'languages', active: ['en', 'tr'], full } — the room's caption languages, to list firstif (msg.type === 'languages') setOpenLanguages(msg.active, msg.full);});
Captions arrive as reliable data messages on the cyron.captions topic, sent only to the participants reading that language. Each utterance has an id: its text grows with updates (final: false) until the final caption (final: true). A status message tells the participant whether their captions are on, or why not: limit (the room already has as many caption languages as the plan allows), unsupported, quota or unavailable.
Every participant also gets a languages message, when it changes and when they join: the caption languages on in the room (first come, first served) and full: true once the plan's limit is reached. Show those languages first in your language picker: anyone can read one of them at once. Another language waits until one of them goes off, then starts by itself (its readers get status active).
Caption minutes are the time speech is transcribed (from when someone starts speaking until a few seconds after they stop), counted for each caption language separately against your plan's monthly caption minutes.
Call length limit
If your plan limits the length of a call, the room is closed that long after it was created, and everyone is disconnected with ROOM_DELETED. In the last 5 minutes the room receives a data message on the cyron.call-limit topic (about every 15 seconds); listen for it to show the time left. Joining the room again starts a new call.
room.on(RoomEvent.DataReceived, (payload, _participant, _kind, topic) => {if (topic !== 'cyron.call-limit') return;const { remainingSeconds } = JSON.parse(new TextDecoder().decode(payload));showCountdown(Date.now() + remainingSeconds * 1000); // the room closes at 0});
SDKs & features
Every official LiveKit SDK works: JavaScript, React, Swift, Android, Flutter, React Native, Unity, Go, Python, Rust and Node.
| Feature | Availability |
|---|---|
| Video & audio rooms, simulcast, screen share | Available |
| Data messages & RPC | Available |
| Server API (RoomService) | Available |
| Webhooks | Available |
| Recording (Egress, to a file) | Available |
| Agents (LiveKit Agents on your servers) | Available |
| Live captions and translation (Scale, Enterprise) | Available |
| Live streaming and ingress (RTMP, HLS, WHIP) | Coming soon |
| Telephony (SIP) | Coming soon |
| Hosted agents | Coming soon |
Service status
Privacy & support
Media flows directly between clients and Cyron's media servers and is never recorded unless you use egress. We store only room and participant metadata needed for usage reporting.
Questions or need a higher limit? Write to [email protected]
