Captions are read in a glance, so the right length depends on where the video plays.
Vertical video (TikTok, Reels, Shorts)
The frame is narrow and the caption usually sits in the middle of the screen, in big bold type. Two to four words per cue keeps each caption on one or two lines and in rhythm with the speaker. Many creators go uppercase for extra punch.
Horizontal video (YouTube, courses, film)
Classic subtitles sit at the bottom and may use two lines of up to about 42 characters. Six to ten words per cue is comfortable to read without rushing the viewer.
Word by word
One word per cue is too fast to read as subtitles, but perfect as data: karaoke highlights, kinetic typography and animated captions use each word's exact timing.
With the Audio to SRT converter you can switch between the three styles after transcribing — the timing stays, only the cut changes.