OpenStudio
github.com
- Free plan
- Web
- API
- Self-hosted
What OpenStudio is
OpenStudio is a free, self-hosted web studio for live call-in shows, multi-host podcasts, and community radio. Peer-to-peer WebRTC mesh audio with mix-minus monitoring supports around 15 participants. Operators can send streams through Icecast to unlimited listeners on their own domain, with unlimited airtime when self-hosted. The browser records separate voice tracks and a program mix, exporting WAV or WebM; MP3 export requires ffmpeg. Optional tools provide on-device whisper.cpp transcription, audio cleaning, and LLM-generated show notes. OpenAI-compatible endpoints include LM Studio, Ollama, OpenAI, Together AI, and Groq; Anthropic needs a compatibility shim. The project lists JWT room tokens, invite tokens, role controls, rate limits, and security headers, and says audio remains on operator-controlled infrastructure. Setup uses git clone, npm install, and npm start at localhost:6736.
Who it suits
OpenStudio suits creators, community radio hosts, and podcast collectives that can operate a self-hosted web application. It is intended for productions needing separate tracks, local recording, and Icecast distribution.
Good to know, and what to watch
- Supports around 15 WebRTC participants
- Unlimited listeners through Icecast
- Separate voice tracks and program mix
- On-device transcription option
- Self-hosted broadcasts have unlimited airtime
- MP3 export requires ffmpeg
- Live demo broadcasts expire after 15 minutes
- Hosts manually share room URLs
- Optional tools require additional dependencies
Plans
Every plan and priceFree / self-host
Free plan
The verdict
OpenStudio provides a self-hosted browser workflow for collaborative audio, streaming, recording, and post-production. Teams should account for its manual invite process and setup requirements.
What the maker publishes
- Audience
- The project describes itself as built for creators, including community radio hosts and podcast collectives. source
- Audio
- It uses peer-to-peer WebRTC mesh audio with mix-minus monitoring and supports around 15 participants. source
- Demo limit
- Broadcasts on the live demo automatically expire after 15 minutes; self-hosted broadcasts have unlimited airtime. source
- Free plan
- Yes source
- Integrations
- Show notes can use OpenAI-compatible endpoints, with examples for LM Studio, Ollama, OpenAI, Together AI, and Groq; Anthropic requires a compatibility shim. source
- Known limitation
- The host interface does not yet expose a button to create invite links, so hosts share the room URL manually. source
- Post-production
- Optional tools provide on-device transcription with whisper.cpp, audio cleaning, and LLM-generated show notes. source
- Privacy
- The README says the audio stays on infrastructure the operator controls and characterizes the project as having zero tracking. source
- Purpose
- OpenStudio is a self-hosted virtual broadcast studio for live call-in shows, multi-host podcasts, and community radio. source
- Recording
- It records separate voice tracks and a program mix client-side, with WAV or WebM export and MP3 export when ffmpeg is installed. source
- Requirements
- Optional transcription and audio-cleaning setup requires ffmpeg, ffprobe, cmake, and about 1.5 GB of free disk space for the default Whisper model. source
- Security
- The project describes 24-hour JWT room tokens, scoped invite tokens, role-based access, WebSocket message caps, per-IP connection limits, rate limits, CORS controls, and hardened security headers. source
- Setup
- The quick start uses git clone, npm install, and npm start, then opens the app at localhost:6736. source
- Streaming
- It streams through Icecast to unlimited listeners using the operator's own domain. source
Checked against the maker's pages on 5 Oct 2026.