MediaSFU is a developer platform for WebRTC video, audio, interactive live streaming, SIP/PSTN calling, realtime AI agents, recording, and live translation. Start with a complete room, embed a focused widget, or use the headless SDK contracts beneath an interface you own.
Website 路 Developer docs 路 Live demos 路 API sandbox 路 Widgets 路 Pricing
| I want to... | Best starting point |
|---|---|
| Open a working room in my framework | MediaSFU QuickStart Apps |
| Add a complete room UI | Choose an SDK and begin with its supplied room |
| Build my own room interface | Headless and custom UI guide |
| Add click-to-call, meeting join, or an agent to a website | @mediasfu/widgets |
| Build a voice or vision agent | MediaSFU Agents |
| Connect SIP trunks, phone numbers, and human takeover | MediaSFU VOIP |
| Run my own MediaSFU media server | MediaSFU Open |
| Explore complete product ideas | Runnable starters and reference apps |
These repositories show how MediaSFU fits inside a real product rather than stopping at a generic meeting screen. Each repository states which platforms were built or exercised; check its platform matrix before treating one target's evidence as proof for another.
| Project | What you can build | Platforms and approach |
|---|---|---|
| QuickStart Apps | The shortest route from SDK installation to a secure room | Web, mobile, Android, Unity, and Apple reference paths |
| Familiar Calls | Contact-first, WhatsApp-style audio and video calls with invitations, draggable mini video, and app-owned call screens | Multi-SDK headless examples plus SDK-free WHIP/WHEP paths and one backend |
| Live Auction | Host video, timed lots, viewers, promotion, reactions, and retry-safe bidding | Multi-SDK custom UI with a shared auction backend |
| Watch Together | Synchronized programs, optional conversation floor, WebRTC delivery, and HLS audiences | React reference plus framework and native integration paths |
| Agents | Voice, vision, knowledge, ordered audio playback, and human handoff | Headless agent starters across web and mobile SDKs |
| VOIP | Browser and mobile calling, SIP/PSTN, agent assistance, and human takeover | Web, mobile, and desktop reference clients with a backend credential boundary |
| SpacesTek Initial | Learn the application flow for social audio/video spaces | Introductory product and navigation layer |
| SpacesTek Final | Secure create/join, local and remote media, roles, and screen sharing | Multi-framework custom room reference |
| SpacesTek Advanced | Headless rendering, moderation, join approval, and request-to-speak | Production-shaped multi-framework architecture |
You can also open the corresponding live showcases for auctions, familiar calls, broadcast control, classrooms, hybrid stages, and watch-together experiences.
MediaSFU maintains ten platform SDK paths plus a shared TypeScript engine. Use the SDK index for current versions, exact installation commands, supported operations, and platform-specific evidence.
| Platform | Package or repository | Best fit |
|---|---|---|
| React | mediasfu-reactjs 路 GitHub |
Web rooms, custom components, UI overrides, and headless products |
| Angular | mediasfu-angular 路 GitHub |
Angular applications and service-owned room integration |
| Vue | mediasfu-vue 路 GitHub |
Vue applications and custom web experiences |
| React Native | mediasfu-reactnative 路 GitHub |
Bare React Native mobile applications |
| Expo | mediasfu-reactnative-expo 路 GitHub |
Expo development builds for iOS and Android |
| Flutter | mediasfu_sdk 路 GitHub |
Mobile, web, and desktop applications from Dart |
| Android | mediasfu-sdk-android |
Android-only Jetpack Compose applications |
| Kotlin Multiplatform | mediasfu-sdk |
Shared Kotlin room logic with supported platform targets |
| Swift and Apple | MediaSFU Apple SDK | Native iOS, iPadOS, and macOS applications |
| Unity | MediaSFU Unity SDK | Games, simulations, training, and scene-owned media |
| Shared core | mediasfu-shared |
Framework wrappers and lower-level TypeScript integration |
The Android and Unity SDKs use MediaSFU-maintained mediasoup transports: mediasoup client for Android and mediasoup client for Unity.
Install the complete package:
npm install @mediasfu/widgetsOr load the production browser bundle:
<script src="https://cdn.mediasfu.com/v1/widget.js" async></script>| Element | Use it for |
|---|---|
<mediasfu-call-button> |
Browser-to-phone, SIP, or support entry points |
<mediasfu-meeting-join> |
Secure room creation and joining |
<mediasfu-ai-agent> |
Voice and multimodal agent sessions |
<mediasfu-web-agent> |
Text, voice, multimodal support, and escalation |
<mediasfu-calls> |
Authorized call operations and history |
<mediasfu-agent-dashboard> |
Agent monitoring and operator takeover |
Import a modular package subpath when you need only one element. React
composition and semantic headless controls are available through
@mediasfu/widgets/react and @mediasfu/widgets/headless.
Package guide 路 Widget platform 路 Widget Studio
- MediaSFU Cloud provides hosted room, media, recording, agent, telephony, translation, and related platform services.
- MediaSFU Open is the MediaSFU media server you run on infrastructure you control. It is not another hosted endpoint and must already be installed and running before an application can connect to it.
Both paths use the same product principle: keep reusable room/API credentials behind an authenticated backend. During isolated local development, use only restricted, revocable credentials and never commit them to a web bundle, mobile app, example URL, log, or screenshot.
Read the secure create/join proxy guide before distributing an application.
- Interactive video rooms, meetings, webinars, classes, and town halls
- Microphone, camera, screen sharing, participant roles, chat, polls, and whiteboards
- Prebuilt room UI, targeted overrides, custom components, and headless rendering
- Cloud recording, pause/resume workflows, transcripts, notes, and translation
- Interactive WebRTC contributors with scalable playback paths such as HLS
- SIP/PSTN calling, browser softphones, IVR flows, agents, and human takeover
- Voice and vision agents with application-owned provider and orchestration choices
- Web, mobile, desktop, Apple, Android, Unity, and framework-wrapper paths
Feature support and runtime evidence differ by SDK and target. Use the operation directory instead of inferring one platform's support from another.
| Resource | Link |
|---|---|
| Developer documentation | mediasfu.com/docs |
| SDK selection | SDK guides |
| Build by outcome | Operations and recipes |
| API experimentation | Interactive sandbox |
| Component previews | Storybook |
| AI agents | agents.mediasfu.com |
| Telephony | Cloud phone and SIP |
| Community support | MediaSFU forum |
| Video guides | YouTube |
| Product updates | LinkedIn 路 X |
Choose the repository that owns the SDK, starter, or product surface you want to improve. Read its contribution and validation guidance, keep credentials out of commits, and describe the exact platform and runtime evidence behind a change. For general platform questions, use the community forum; for a reproducible repository defect, open an issue in that repository.
Build the workflow, not the realtime plumbing.
Start free 路
Try a live demo 路
Open the docs