Content and display are separate questions
Subtitles generally convey dialogue, sometimes in another language. Captions can also describe information a listener gets from sound, such as who is speaking, music or a meaningful sound effect. Closed describes a switchable track; open describes text in the picture.
Decide what your audience needs
A dialogue-only transcript may omit a doorbell, off-screen voice or music cue that matters to understanding the scene. Add that context when it helps the audience. Avoid filling the screen with descriptions that add no useful information.
Choose a file format after the content
SRT and VTT can carry dialogue and descriptive text. Neither extension guarantees that a track is complete, accurately timed or accessible. Review the information against the whole video, not only the speech transcript.