How to transcribe a podcast episode to text

Podcasts are audio-only by nature, which is exactly why they're hard for anyone — listeners, search engines, or you six months from now — to search, quote, or skim. A transcript fixes that. Here's why it's worth doing, and how to actually turn an episode into text.

Why podcasters bother transcribing at all

A few concrete reasons this is worth the effort beyond "nice to have":

Search engines can't listen to audio. An episode with no transcript is invisible to text search — Google has nothing to index but your title and description. A full transcript gives every name, term, and topic you discussed a chance to rank on its own.
Show notes write themselves faster. Writing show notes from memory after recording is slow and you forget things. Writing them from a transcript is mostly editing and trimming, not composing from scratch.
Chapters and timestamps. A timestamped transcript makes it easy to mark where topics start, which is what podcast chapter markers are built from.
Accessibility. A transcript makes the episode usable for deaf and hard-of-hearing listeners, and for anyone who'd rather skim than listen.
Repurposing. Clips, quote graphics, newsletter excerpts, and blog posts pulled from an episode all start from having the words in text form.

None of this requires anything fancy — it requires an accurate transcript that exists in the first place.

Getting a transcript: audio URL or uploaded file

Most podcast episodes exist as an audio file, either hosted somewhere with a direct URL (your host's CDN, an RSS enclosure link) or sitting on your machine after you've exported it from your recording software. Either starting point works for transcription — you don't need video, and you don't need the episode to be published anywhere first.

Doing it yourself, for free

Reasonable if you're transcribing occasionally or on a tight budget:

Free transcription tools with a minutes or file-size cap. These usually cover a single short episode, though most podcast episodes run well past typical free-tier limits, so this stops working once you're doing this regularly.
Manual transcription. Realistic for a five-minute clip you want to quote; not realistic for a 45-minute episode — expect several times the episode's length in listening and typing.
Run-it-yourself speech-to-text, if you're comfortable with a bit of setup: pull the audio, run it through a speech recognition model, format the output. Workable, but you're now maintaining a small pipeline — chunking long files, handling failures, storing the audio — for something that isn't your actual job.

These are legitimate options for a single episode or an occasional need. Nobody running one show a month needs a subscription for that.

Where a dedicated tool earns its keep

Once transcription is a recurring part of publishing — every episode, every week — the free-tier caps and manual typing stop being practical, and a DIY pipeline becomes ongoing maintenance work instead of a one-time script. That's the case for a tool that takes either the episode's audio URL or an uploaded file and returns a finished transcript without you managing the speech-to-text step yourself.

Using transcript.land

transcript.land accepts a direct audio URL or an uploaded file, so it works whether your episode is published with a hosted link or you're transcribing from a local export before you publish anything.

Shell

curl -X POST https://api.transcript.land/v1/transcript \
  -H "Authorization: Bearer $TRANSCRIPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://your-host.com/episode-42.mp3", "format": "json"}'

Or from the terminal:

Shell

brew install ziqorg/tap/transcript
transcript login
transcript get "https://your-host.com/episode-42.mp3"

The transcript comes back with timestamps, which is what you need for chapter markers, and exports as TXT for show notes copy, SRT or VTT if you're publishing a video version of the episode and want captions, Markdown for a blog writeup, or JSON if you're pulling it into another tool. The same workflow also handles video-hosted episodes, since it works with YouTube, TikTok, X, Instagram, Bilibili, Facebook, and RedNote links too, so video podcasts don't need a separate process.

There's a free tier — 15 minutes a month — enough to transcribe a short episode or try it on a past one before committing. The paid plans (currently a $19/month tier with 20 hours and a $99/month tier with 200 hours) are built for a regular publishing schedule rather than one-off use. It isn't the only transcription service built for podcasts, and it isn't positioned as the cheapest option available — the value is one workflow that handles an audio URL, an uploaded file, or a video link the same way, with timestamps and export formats meant for actual show notes and subtitle work, not just a wall of text.

Turning the transcript into show notes and chapters

Once you have the timestamped transcript:

Trim it into show notes. Pull the transcript into your editor, cut the filler and false starts, and keep the structure — this is usually faster than writing notes from memory.
Mark chapter points. Scan the timestamps for topic changes and use those as your chapter markers; most podcast hosts accept a simple timestamp-plus-title format.
Pull quotes for promotion. A transcript makes it easy to find and copy the exact line you want for a social post or newsletter excerpt.

Summary

Transcripts make podcast episodes searchable, give you a starting point for show notes and chapters, and support accessibility and repurposing.
Free tools and manual typing work fine for an occasional single episode.
For a recurring publishing schedule, transcribing directly from an audio URL or an uploaded file with a tool like transcript.land beats running your own speech-to-text pipeline for something that isn't your core product.