Most third-party captioning services will provide well-formatted captions. If you're reviewing captions, or are creating captions from scratch, here are a couple of tips to keep in mind:
Identify speakers: With multiple speakers (especially with an offscreen interviewer), identify who is speaking by adding their name in brackets with a colon: [Interviewer:]. You can use just the last name after initially identifying the person by their full name.
Do transcribe exactly what the person is saying, including grammatical errors or accents: "y'all", not "you all" for example.
Skip filler words: You do not need to transcribe "ums", "uh" or other filler words.
Use punctuation, capitalization, and grammar: Use the same language mechanics as if you were writing a text document.
When a person quotes another, use quotation marks.
Use an ellipsis (…) if there is a significant pause.
Use italics if the speaker emphasizes a word.
Include non-speech sounds: Add captions for background sounds and music when they are important for understanding the context of what's happening in the video. Examples: [Door creaking], [Dramatic music].
Keep captions short: Each caption should be no longer than 2 lines and should not obscure any important information in the video. Aim to keep complete sentences and phrases together and don't split them across screens.