Small media tools should be useful without asking for your files.

CaptionKite is an independently built collection of browser-based tools for everyday video work. It began with a simple problem: creating a caption file should not require opening an account, buying a monthly plan, or uploading a private recording to an unfamiliar server. Modern browsers can now decode media, run speech-recognition models, trim clips, extract audio, and build animated images. When the browser can safely do that work on your own device, we believe it should.

Why the tools work in your browser

With a conventional online converter, your file travels across the internet, waits in a processing queue, and is stored on somebody else's computer while the job runs. CaptionKite takes a different route. A local file is read by the page, processed with browser technologies such as WebAssembly or WebGPU, and returned to you as a download. The caption generator also downloads its speech model to your browser, where it can normally be cached for later visits.

That design reduces unnecessary exposure for interviews, meeting recordings, lectures, family videos, and work in progress. It also avoids the bandwidth bill that forces many free converters to impose tiny upload limits or short daily allowances. CaptionKite does not need an account to connect your jobs to a profile, and it does not put a watermark on the result.

The honest tradeoff

Local processing moves the work from a remote server to your computer. A recent desktop with Chrome or Edge will usually handle more than an older phone or a machine already short of memory. Long recordings take longer than short ones. Codec support differs between browsers, and a file extension such as MP4 does not guarantee that the audio and video inside can be decoded. The limits shown beside each tool are technical ceilings, not a promise that every device can process a file at that maximum.

We explain those constraints because privacy claims are not useful if the tool quietly falls back to an upload. CaptionKite does not use a hidden server-side conversion path. If a browser cannot preview or decode a file, the page reports the problem and suggests a practical alternative.

What CaptionKite deliberately does not do

The site does not bypass passwords, viewing permissions, disabled downloads, or access controls. A normal YouTube, Vimeo, or Zoom viewing page is not the same as a direct media file, and CaptionKite will not pretend otherwise. You should only process recordings you own or are authorised to use. We also avoid claiming that automatic captions are a certified record: names, numbers, specialist terms, and difficult audio always deserve human review.

Built in public, improved through real use

The tools are maintained by a small team that reads its own bug reports. Bug reports often reveal the cases that matter most: a phone recording with an unusual orientation, a WebM without a normal duration field, an unfamiliar codec, or a long lecture that stresses browser memory. If something is unclear or fails, use the contact form. Include the browser, operating system, file type, approximate duration, and exact message you saw. Please do not send confidential media, passwords, or private access links.