Hardcoded captions, made on your machine.
This plays your video once, in real time, drawing each caption onto the picture as it goes — so exporting a 90-second clip takes about 90 seconds, the same way Auto Subtitle’s transcription time tracks the clip’s own length. There is no way to record faster than the video plays.
The picture is also re-encoded in the process, not copied — unlike Video Muter or Video Trimmer, which never touch a pixel, this is a genuine generational step. Keep your original file if you might need it uncaptioned later.