How havenConverter works
havenConverter is fully client-side. When you open a video, the file is read from your disk into the browser tab — not sent to a server.
1) Open
Drag and drop or browse for a video. The file is demuxed in place (with Mediabunny) to read its codec, resolution, duration and audio track. Nothing is transmitted over the network.
2) Edit
Two engines share the work. Engine A uses the browser’s own WebCodecs decoder and encoder with a WebGPU scaler — that is the fast, hardware-accelerated path for transcoding, compressing and resizing, and for the multi-track timeline. Engine B is a self-hosted ffmpeg.wasm build that handles everything the platform can’t: legacy containers such as AVI, concatenation of mismatched clips, watermark overlays and audio mixing.
3) Export
The output is muxed in the browser and streamed to a file as it is produced, so multi-gigabyte 4K exports keep memory flat. Save to your downloads or pick a folder — the file is written straight to your device.
Honest acceleration badges
Every job shows whether it ran on hardware, software or the CPU (wasm) path. The badge is derived from the browser’s Media Capabilities answer for that exact codec and resolution, not guessed.
What you need
| Browser | Engine A (WebCodecs + WebGPU) | Engine B (ffmpeg.wasm) |
|---|---|---|
| Chrome / Edge | Yes, hardware decode and encode where available | Yes |
| Firefox | Yes | Yes |
| Safari | Yes | Yes |
You can run the same check on your own machine at /selftest — it performs a real transcode with both engines and reports pass or fail.
Works offline
After the first visit the studio and its processing engines are cached by a service worker, so havenConverter keeps working without a connection and can be installed as an app.
Open havenConverter →