Privacy and How It Works

Your videos never leave your device

When you pick a video, the browser reads it from your device and the conversion runs in this page, using your browser's built-in video encoders (WebCodecs, through the Mediabunny library) or FFmpeg compiled to WebAssembly. The finished file is created on your device and saved through your browser's normal download. The video's contents, its file name and any watermark image you add are never uploaded to this site or to anyone else.

The tools that use AI models also run on your device. Blur Faces finds faces with YuNet, Remove Video Background finds the person with MODNet and MediaPipe, and Auto Caption Video transcribes speech with Whisper. These tools download model files to your browser. No video frames, audio or transcripts are sent anywhere.

Once the page has loaded, conversions that use the browser's own encoders keep working without an internet connection. The FFmpeg engine, used for some formats and older browsers, is downloaded the first time a file needs it, which can be as soon as you pick the file: reading an AVI, WMV or WMA file, or making its preview, already uses FFmpeg. Your browser keeps it after that, so it works offline from then on.

Some files are prepared before the conversion: FFmpeg repacks a container the browser cannot read, decodes audio it cannot play, or decodes the video once into an intermediate copy, and the preview may get a small copy made the same way. These steps run on your device like everything else; the prepared copies stay in your browser's memory and are discarded when you remove the file or close the tab.

What the app downloads

• The app itself (pages, scripts, styles and fonts) from privatevideoconverter.com. Mediabunny is part of these scripts.
• The FFmpeg engine files (a WebAssembly file of about 30 MB), also from this site, only when a file needs FFmpeg: when you pick a file the browser cannot read or decode on its own (to inspect it and make its preview) or when a conversion needs it. Your browser keeps a copy so it is not downloaded again.
• ONNX Runtime Web, MediaPipe and the AI models (YuNet, MODNet, the MediaPipe selfie segmenter, Silero VAD and Whisper), also from this site, the first time you use the face detection, background removal or caption tools. The caption models are the largest: Whisper tiny is about 44 MB, base 80 MB and small 252 MB, or 122 MB, 209 MB and 589 MB when your browser runs them with WebGPU, plus 29 MB once for the speech engine. Only the model you choose is downloaded. Your browser keeps a copy.
• App updates. A service worker stores the app for offline use and checks for new versions; you are asked before an update is applied.

These are ordinary downloads from this site to your browser. Like any website, the server and its network provider (Cloudflare) can see your IP address and which pages and files are requested.

What Google Analytics records

The site uses Google Analytics 4 to see which tools people use and where conversions fail. It receives:

• the page you are on, and the campaign tag in the link that brought you here, if any;
• which tool ran and its settings (output format, codec, quality, resolution, which caption model);
• file sizes and durations, grouped into ranges, never the exact value;
• when Video Info successfully reads a file, its detected container, video and audio codecs, resolution, and size and duration ranges;
• how long a conversion took, which engine ran it, and an error code when it fails or is cancelled;
• whether you scrolled to the bottom of a page, and clicks on links to other sites.

It never receives your files, their names, the tags found in them, the areas you blur, the faces found or the text of your captions. As on any site that uses Google Analytics, Google also sees your IP address and basic browser and device details. Google signals and ad personalisation are turned off. See how Google uses information from sites that use its services.

What is stored on your device

• Your settings and saved presets, in your browser's local storage.
• The app files, and the FFmpeg engine, ONNX Runtime, MediaPipe and AI model files once a tool has used them, in the browser cache, so the tools work offline.
• Nothing else from your videos. Outputs are held in memory until you download them. Large outputs can be written straight to a file you choose (desktop Chrome and Edge) instead of being held in memory.

None of this is sent anywhere. Clearing this site's data in your browser settings removes it.

Tools for sharing a video privately

• Prepare Video for Sharing trims, removes audio, strips identifying tags, blurs or covers areas and fits a size limit in one pass, then checks the result.
• Remove Video Metadata removes GPS location, device and date tags without re-encoding the video.
• Blur Faces in Video finds faces automatically and blurs, pixelates or covers them.
• Blur Video blurs, pixelates or covers areas you draw, for the time range you choose.
• Remove Video Background replaces what is behind a person with a colour, an image or a blur.
• SpectrShare sends a video straight from your browser to the recipient's browser over WebRTC, with no upload to a server. It is a separate site also created by me.