How to Transcribe a Podcast on Your Mac (Locally, in Seconds)
A practical guide to transcribing podcasts on your Mac without uploading anything — and what to actually do with the transcript once you have it.
The transcription takes twelve seconds. The interesting part is what you do with the text afterward — and that’s where most guides stop.
What you need
- A Mac with Apple Silicon (M1 or later)
- macOS 14.2 or later
- MacParakeet (free download — free and open-source)
- A podcast file (MP3, M4A, WAV, MP4, or any common audio/video format) or a YouTube URL
Getting the podcast file
If you have the file, skip ahead.
Apple Podcasts: Download the episode, right-click it, select “Show in Finder.” Apple buries these in nested directories — the right-click saves you from hunting.
Podcast websites: Look for a download icon near the player, or right-click the play button and choose “Save Audio As.”
RSS feeds: Open the feed URL in a browser, search for <enclosure url= — those are direct audio links. Copy and download.
YouTube: Paste the URL directly into MacParakeet — it downloads the audio and transcribes it automatically. No extra tools needed.
The transcription
Drag the file into MacParakeet, or paste a YouTube URL.
| Episode length | Time |
|---|---|
| 15 minutes | ~3 seconds |
| 30 minutes | ~6 seconds |
| 1 hour | ~12 seconds |
| 2 hours | ~24 seconds |
| 3 hours | ~36 seconds |
Everything processes on your Mac. No upload, no cloud, no queue.
Command line
macparakeet-cli transcribe episode.mp3Output to stdout. Pipe to a file:
macparakeet-cli transcribe episode.mp3 > transcript.txtBatch a full season:
for f in episodes/*.mp3; do
macparakeet-cli transcribe "$f" > "transcripts/$(basename "$f" .mp3).txt"
doneWhat to do with the transcript
A text file in a folder isn’t useful. Here’s where transcripts earn their keep.
Write show notes in minutes
If you produce a podcast, show notes are the most tedious part of publishing. You either re-listen to the full episode, work from memory, or skip them entirely.
A transcript lets you skim-read an hour of conversation in ten minutes. Every topic, guest quote, and resource mention is searchable text you can highlight and copy from. Hand it to an LLM for a draft, then edit. Either way, the transcript is raw material that makes everything downstream faster.
Make your back catalog searchable
The most underrated use of podcast transcription, and it compounds over time.
Once transcribed, every episode becomes searchable text. “What episode discussed pricing strategy?” Search for “pricing.” “When did the guest mention that sleep study?” Search for “sleep.” Instant results across years of content previously locked inside audio files.
The batch command above makes this practical — two hundred episodes averaging 45 minutes each is 9,000 minutes of audio, which processes in about 30 minutes at 300x realtime. The archive is useful forever.
Pull exact quotes
Podcast episodes are dense with quotable moments — but nobody re-listens to an hour of audio hunting for them. A transcript lets you copy exact words for social posts, newsletters, or audiogram captions.
Cite and reference
Journalists, academics, and researchers need exact quotes, not paraphrases from memory. A transcript is a citable source document. An hour of audio becomes a reference text in twelve seconds.
Tips
Audio quality is the biggest variable. Two people with decent microphones in a quiet room — Parakeet handles this easily. Accuracy drops with heavy background music, street noise, extremely low-bitrate audio (below 64kbps), and phone-quality recordings. If you’re transcribing your own podcast, a $50 USB mic pays for itself in both listener experience and transcription accuracy.
Accents work well. Parakeet TDT handles British, Australian, Indian, Southern American, and other English accents with high accuracy. Strong regional dialects may produce occasional errors, but performance stays solid across the range.
Speaker diarization built in. MacParakeet detects and labels individual speakers automatically. Rename speakers inline (e.g., “Speaker 1” to “Alice”) and see per-speaker stats. Labels carry through to all export formats.
Length doesn’t matter. No duration limit, no quality degradation. Three hours takes about 36 seconds.
Why local for podcasts
Cloud transcription works fine for podcasts. But local is specifically better in a few cases:
Unreleased episodes. No pre-release audio on third-party infrastructure.
Sensitive interviews. Off-the-record conversations stay on your machine.
Back-catalog economics. Cloud services charge per minute. Two hundred episodes can cost hundreds of dollars. MacParakeet is free and open-source.
No connectivity required. Planes, cabins, road trips.
Try it
Download MacParakeet, drop in a podcast file or paste a YouTube URL, and see what twelve seconds looks like. It’s free and open-source. Transcribe your entire back catalog if you’re motivated.