Captions

What actually happens to your recording

Speech recognition here is done by Whisper, running as a model inside your browser rather than as a service somewhere else. The first time you use one of these tools the model weights download and are cached, which takes a moment; after that a transcript is produced entirely on your machine. This matters more for audio than for anything else on the site. A recorded meeting, a therapy session, a legal interview or a doctor's dictation is exactly the kind of file that should not be handed to a transcription service, and here it is not handed to anyone.

Which tool for which job

Transcribe gives you the words, as text or as a subtitle file you can edit elsewhere. Auto captions goes further and burns styled, readable captions into the picture, which is what a video needs for the platforms that autoplay muted. Burn subtitles is for when you already have an SRT or VTT and simply want it permanently in the frame. If you are assembling something short for social, the short maker and the editor both include captioning as part of the edit.

Read the transcript before you ship it

Whisper is good and is not correct. It struggles with proper nouns, technical vocabulary, heavy accents, crosstalk and anything recorded at a distance from the microphone, and it will occasionally produce a fluent sentence that was never said. Every tool here shows you the transcript in an editable form before it goes anywhere near the video, which is not a formality: on any recording that matters, read it through. Clean audio helps far more than any setting does, so if you have the original rather than a compressed copy, start from that.