recorder.video

recorder.video is for the moment you need to show someone something on your screen and don't want to install software, make an account or hit a time limit halfway through the demo. You pick what to record, a whole display, one window, a single browser tab, your webcam, or your screen with your face in a bubble, and the browser asks for the matching permission. Your microphone can come along, and on some setups so can tab or system audio.

While you record, each take is written to the browser's storage in small pieces, so a crash or an accidental reload doesn't lose it. Afterwards you can trim the clip and export it as MP4, WebM, audio only or a short GIF. The conversion runs on your own machine with the browser's WebCodecs and the open-source mediabunny library.

Background blur for the webcam uses MediaPipe's selfie segmentation model, served from the site and run locally. A teleprompter keeps your script on screen, drawing tools let you mark up what you're showing, and size presets fit YouTube, Shorts and TikTok. Captions are optional. Turn them on and the app downloads the open-source Whisper speech model, about 75 MB, from Hugging Face once, then transcribes on your device. Guides cover recording on Mac, Windows and Chromebook, and there are separate webcam and voice recorders.

The privacy page says recordings are stored on your device and never uploaded, and there's no upload path in the app at all. There are no accounts and no ads. Google Analytics measures visits, and you can turn its cookies off in Cookie settings. Hugging Face only sees the model download request if you turn captions on.