Captions are a synchronised transcription (or translation) of all meaningful audio in a video, including music, sound effects, laughter, applause, etc.
Captions assume that the viewer cannot hear the audio at all. Captions are less commonly translated compared to subtitles; however, translated captions are becoming increasingly popular for social media content, since videos are often muted in feeds.
Captions also include much more information than subtitles, as their aim is for the viewer to be able to absorb just as much information from reading the captions as they could by listening to the audio.