How to make a podcast transcript on a Mac (and why publish it)

Last updated 21 October 2026

A podcast transcript is the text of an episode, usually with timestamps. Hosts want one for show notes and chapters; listeners want one for accessibility and skimming; search engines can only index text. Getting one on a Mac is a matter of running the finished episode, or the raw recording, through a speech model.

This guide covers the three routes (macOS, a hosted uploader, or transcribing on the Mac), what makes podcast audio easier or harder than other recordings, and what to do with the transcript once you have it.

Which file to transcribe

If you're the host, you have a choice that listeners don't: transcribe the mixed master, or the separate recordings of each person.

The mixed master: one file, one pass, and the transcript matches what listeners hear, including edits. The drawback is that a single mixed track carries no reliable record of who said what; a speech model transcribes words, not identities, so you'll label speakers by hand.
Per-speaker stems: if you recorded each guest on their own track, transcribe each one separately. Each transcript is then one voice, labelled for free, and timestamps line the two up. Crosstalk and interruptions also come out cleaner.
A stereo master with one person per channel: split it into two mono files first; you get the per-speaker benefit without needing the stems.

If you're a listener wanting a transcript of someone else's show, you have the published file only. Check whether the publisher already provides a transcript before making your own, and respect their licence if you plan to publish the text.

What macOS can and can't do here

Apple's built-in transcription (Voice Memos and Notes) transcribes recordings those apps made themselves. Voice Memos does it on macOS Sequoia with Apple silicon; Notes on Macs with M1 or later in a listed set of device languages. Neither imports an MP3 episode, so for a podcast file they don't apply unless you recorded the episode in Voice Memos in the first place. Live Captions will caption an episode while it plays, on device, but it's a live window rather than a saved transcript, and Apple notes accuracy may vary.

Uploading the episode to a hosted service

Otter's MP3-to-text page says to log in or create a free account, then upload the file. The limits stated on its pages matter for podcasts specifically: the free plan is listed at 300 transcription minutes a month, files up to 90 minutes, and 3 pre-recorded file imports per user, lifetime, according to the pricing page. A weekly show runs out of that in a month. Pro is $16.99 per user per month ($8.33 billed yearly), with 10 imports a month and 1,200 imported minutes. Transcription languages are English (US and UK), Spanish and French, one at a time, and exports are TXT, DOCX, PDF and SRT.

The trade: the episode is uploaded and the transcript lives in the service's account. For a published show that's rarely a problem. For unreleased episodes, interviews recorded under embargo, or guests who asked for the raw recording not to be shared, it can be.

Transcribing the episode on the Mac

QuietNote is a note taker for Mac. Import a common audio or video file and it's transcribed on your Mac with the Whisper model that ships with the app, so it works offline and nothing is uploaded. The output is a Markdown transcript with timestamps in a folder you choose, which suits a show-notes workflow: the file is plain text, versionable and easy to paste into an episode page. A summary is optional, either with a local model through Ollama or with Claude using your own API key (only the transcript text is sent, only when you ask).

Import the episode, or each speaker's stem.
Wait for the transcript. A 60-minute episode takes a while; how long depends on the Mac and the model size.
Open the Markdown file, fix names and jargon, then build show notes and chapters from it.

There's no per-episode minute cap, because the Mac does the work. The first signed Mac release isn't out yet; leave your email to hear when it ships. QuietNote needs macOS 14.2 or later.

Podcast audio and transcription quality

Good news: podcasts are usually close-miked, edited and in one language, which is the easiest audio a speech model sees. Whisper-class models do well with it.
Names, brands and jargon: these are what a model gets wrong most, and a podcast is full of them. Keep a list of correct spellings and search-and-replace them in the transcript.
Intros, music and ads: speech under music may be partially transcribed. Trim them or expect some noise in those sections.
Laughter, overlap and interruptions: on a mixed track these become garbled lines. Stems avoid most of it.
Accuracy numbers: nobody can promise a percentage for your episode. Whisper's README notes its English-only model variants tend to perform better for English; beyond that, read the transcript before you publish it.

Turning the transcript into show notes, chapters and a page

Show notes: a short summary plus the links mentioned. The summary can come from a local model, then you edit it; the links you pull from the transcript by searching for the sites and books that came up.
Chapters: go through the timestamps and note where the topic changes. A chapter list in the episode description is the most useful thing a transcript gives you for the least effort.
Quotes for social posts: pick lines with their timestamp so you can cut the audio clip to match.
Published transcript: paste the cleaned text on the episode page. Deaf and hard-of-hearing listeners can read it, people can skim before committing an hour, and search engines index what the episode is about, which audio alone can't give them.
Subtitles: if you also publish on video, you need an SRT rather than a transcript; Otter exports one, or take the transcript to a subtitle editor.

Questions

How do I get a transcript of a podcast?

If the publisher provides one, it's usually on the episode page. If not, transcribe the episode file yourself: upload it to a hosted service such as Otter (free plan listed at 300 minutes a month, files up to 90 minutes, 3 lifetime imports) or import it into a local tool such as QuietNote, which transcribes it on the Mac.

How long does it take to transcribe a 60-minute podcast?

Hosted services say roughly the length of the file: Otter's pages give a five-minute file taking about five minutes. On a Mac it depends on the Mac and the model size; recent Apple silicon machines are usually faster than real time. Either way the editing afterwards, fixing names and adding chapters, takes longer than the transcription.

Should I publish transcripts for my podcast?

Yes, if you can. A transcript makes the episode accessible to people who can't listen, lets everyone skim it, and gives search engines text to index. Clean it first: fix names, remove the worst filler, and keep timestamps at chapter boundaries.

Can transcription tell which host is speaking?

Not reliably on a single mixed track; a speech model transcribes words, not voices. Transcribe each speaker's own recording if you have the stems, or split a stereo file with one person per channel into two mono files. Otherwise label speakers by hand using the timestamps.

Can I transcribe someone else's podcast?

You can make a transcript for your own use from a file you have. Publishing it is a different matter: the episode is the publisher's work, so check their licence or ask before posting the text.

Sources

Checked on the “Last updated” date above. Vendors change their settings and prices often; follow the links for the current version.