Work in order. A mistake you fix early (a misspelled guest name, the wrong caption track) otherwise gets copied into every asset that follows.

Wait for the captions before you do anything

YouTube does not caption a video the moment it finishes uploading. Auto-captions are generated after processing, and on a 90-minute interview that can take noticeably longer than on a short clip. Caption-based tools read the tracks YouTube holds; they do not listen to your audio. So if you paste the link too early, you get nothing.

On TranscriptYou, a video with no caption track returns a plain message, "No subtitles or transcript are available for this video", after about two seconds. That message on a fresh upload usually means "come back later", not "this video can't be transcribed". The same applies to private, scheduled and members-only uploads: if the episode is still private while you prepare the premiere, the tool cannot see it.

A practical habit: upload the video the evening before the audio goes out, then pull the transcript the next morning. By then the auto-captions exist and you have the text in hand before the episode hits the feed.

Pick the right track

Open the "Transcript language" menu before you press Download. A video can carry several tracks, and the one selected first is not always the one you want.

If you uploaded your own caption file in YouTube Studio, the episode has at least two English tracks: yours, labelled "English", and YouTube's machine version, labelled "English (auto)". Yours is the better source. It has correct names, punctuation and sentence breaks you already paid for (in time or money). Choose it.

If you never uploaded captions, "English (auto)" is the only option, and that's fine for show notes. Just budget time for the corrections described further down. Shows with guests speaking other languages, or with dubbed or translated tracks, should check the menu every time, since grabbing an auto-translated track by accident produces notes that read slightly off in ways that are hard to spot later.

Fix guest names and jargon once, at the source

Auto-captions are good at ordinary speech and bad at exactly the words your listeners care about: your guest's surname, their company, the product they built, the acronym your niche uses every day. A guest called Siobhan may appear as "Shivon" in one minute and "Chevonne" in the next.

The fastest check takes a minute. Search the downloaded text for the guest's name as it should be spelled. If it appears fewer times than you know they were introduced and addressed, the captions mangled it. Then search for your show's name and your sponsor's name, because those are the two errors that embarrass you in public.

You have two places to fix things. For a one-off, do find-and-replace in your text editor. For anything that will stay public, correct it in YouTube Studio's subtitle editor; the edited version is published as its own track, which improves the captions viewers see and gives you a clean source for any future pull. Next time you open the transcript menu, choose that corrected track instead of "English (auto)".

One more cleanup item: some exported lines keep the mid-line breaks of the original captions. Joining them before you paste text into a CMS saves the episode page from looking like a chat log.

Turn the text into the assets you actually publish

Chapter timestamps

Download the .txt with timestamps switched on. Each line starts with its time, in the form [0:27] Good morning..., so you only need to scroll for topic changes and copy the time from the nearest line. For YouTube chapters, the first timestamp in the description must be 0:00 and you need at least three of them. The same list works in your show notes and in podcast apps that read chapter markers from the description.

Show notes

Write the summary from the transcript, not from your memory of the recording. You will quote the guest correctly, and you will catch the book, tool or study they mentioned in passing that deserves a mention in the notes. Keep this part short: a few sentences on what the episode answers, the chapter list, then whatever the guest asked you to include.

Quote cards

Switch timestamps off for this step so you get clean prose. Read for sentences that make sense without context, roughly one to two lines long. Check each one against the audio before it goes on a graphic. A misheard word in a quote card is the error guests notice first.

Episode page for search

A full, cleaned transcript under the player gives search engines a page of text about exactly what you discussed, including the long, specific questions people type. Add it below the show notes, with the chapter headings reused as subheadings, so a reader can jump to the part they came for.

Accessibility transcripts for the feed

Deaf and hard-of-hearing listeners, people in noisy places and anyone who prefers reading all benefit from a transcript linked from the episode. If your hosting setup supports the Podcasting 2.0 tag, SRT is one of the formats it accepts, alongside plain text and HTML. Check your hosting dashboard for a transcript field before assuming you need to convert anything.

A small catch: TranscriptYou exports .txt and .srt only, with no VTT. If a player or plugin of yours insists on VTT, you will need a converter for that one file.

SRT for clips

For short vertical clips, download the .srt instead of the text. The cues use the standard format with milliseconds, for example 00:00:27,103 --> 00:00:29,678, and both Premiere Pro and DaVinci Resolve import them directly. Cut the clip, import the full episode's SRT, and delete the cues outside your in and out points. Then re-time the first cue to zero, or shift the whole block, depending on your editor.

Before exporting the clip, open the .srt in a text editor and confirm the guest's name is spelled correctly inside the cues. Captions burned into a clip cannot be fixed after it has been shared.

When this workflow doesn't fit

Caption-based tools depend entirely on YouTube having captions. If your show is audio-only, you publish to YouTube days after the feed, or YouTube's auto-captions are weak in your language, you need speech-to-text that runs on your audio file. The same goes for episodes you never make public.

For a video-first show, try it on last week's episode. Paste the YouTube link, check which track is selected, and search the text for your guest's name before you copy a single line.