ShowHN: Caption a video in the browser with Whisper, video not uploaded (stayput.dev)

A new web-based tool allows users to automatically add large, readable captions to videos in their browser without uploading any content. Leveraging OpenAI's Whisper model, this open-source solution operates offline after the initial download of a 76 MB speech model. It generates captions from the audio of MP4, MOV, or WebM videos, embedding them directly into the video frames. This functionality addresses the growing trend of viewing videos on mute across social media platforms, ensuring that videos maintain engagement even when sound is not activated. The tool stands out for its privacy and usability features: it does not require an account, does not impose a watermark on the finished product, and keeps all processing local to the user's device. Users can customize the appearance of the captions and choose where to position them in the frame, maximizing visibility against various backgrounds. Importantly, it facilitates the creation of SRT files for those who wish to refine the captions further. Overall, this innovation enhances accessibility in video content creation, providing a free and efficient way to cater to audiences who prefer or need visual subtitles.