subany.video

Problem

Live streams and videos are often only available in one language, and viewers who need subtitles depend on whatever the platform provides. Sending audio to a third-party service adds latency, cost and a privacy question, while asking people to install an app or create an account adds friction before they can watch anything.

Approach

subany.video adds subtitles and translation directly in the browser. A speech-recognition model runs on the viewer’s own machine with WebGPU, so audio does not need to be uploaded to a server. The recognised text is translated and drawn over the stream or video as it plays. There is no account and no installation step: the site opens, the video or stream is selected, and subtitles appear.

Result

Viewers get live subtitles and translation while the media stays local to their device. The implementation is private, and the step-by-step explanation of how the pipeline works is available as a separate article for anyone who wants the technical detail.