How to transcribe a podcast episode (four methods compared)
Practical comparison of four ways to get a transcript from a podcast episode — with real accuracy numbers, cost, and turnaround time for each method.
Podcast transcription used to mean paying a professional or spending hours typing. Now it takes a few minutes and a few dollars — or nothing if you're on a free plan.
Here's how four different approaches compare, using a 45-minute interview episode as the test case.
Why transcribe a podcast
Three practical reasons:
Search. Audio doesn't rank. Text does. A transcript turns a 45-minute interview into a page that search engines can index. Most podcast episodes cover enough ground to rank for dozens of related queries if the content is there.
Repurposing. A transcript is raw material for a blog post, newsletter, social clips, or a summary. You're not rewriting from scratch — you're editing.
Accessibility. Deaf listeners and people in noisy environments can read what they can't hear. Some podcast platforms now require transcripts for compliance.
Method 1: Upload the MP3 to a transcription tool
The fastest approach for most people.
Download the episode's MP3 (most podcast apps have a download button, or you can find the direct URL in the RSS feed). Upload it to TranscribTxt or a similar AI transcription tool. Wait 4-6 minutes for a 45-minute episode. Download the transcript as TXT or SRT.
What you get: A rough transcript with 94-96% accuracy on clear audio. Speaker changes are marked as paragraph breaks but not labeled by name. You'll need to add speaker labels and fix the occasional wrong word.
Cost: Free for 5 files per month on TranscribTxt. $12/month removes the limit and adds SRT export.
Time: About 10 minutes total, including downloading the file, uploading, and a quick review.
Method 2: Use the podcast's own transcript
Before you transcribe anything, check if a transcript already exists.
Many podcasts publish transcripts on their website, in the episode show notes, or on dedicated transcript sites like Podscribe or Listen Notes. Some publishing platforms (Buzzsprout, Transistor) auto-generate transcripts for each episode.
Search [podcast name] transcript [episode title] — you'll find existing transcripts more often than you'd expect.
What you get: A transcript that's already been reviewed, often with speaker labels. Quality varies depending on the publisher.
Cost: Free.
Time: 2 minutes to check.
Method 3: YouTube captions
If the podcast is also on YouTube, you can get the auto-generated captions.
Open the episode on YouTube, click the three dots below the video, and select "Open transcript." You'll get a time-coded text that you can copy. The accuracy is lower than dedicated transcription tools — YouTube captions don't handle two speakers talking simultaneously well, and they skip over crosstalk.
A quicker workaround: use a YouTube transcript tool (like the one on TranscribTxt) to extract the transcript directly from the YouTube URL without opening the video.
What you get: A usable rough draft, lower accuracy than Whisper-based tools, no speaker labels.
Cost: Free.
Time: 5 minutes, including cleanup.
Method 4: Manual transcription
Typing out what you hear. Professional transcriptionists work at roughly 1 hour of audio per 4-6 hours of typing, faster with foot pedal software like oTranscribe or Express Scribe.
For most purposes, this is only worth doing when:
- Accuracy requirements are very high (legal depositions, medical records)
- Audio quality is too poor for AI (heavy background noise, extremely thick accents)
- You need 100% verbatim with every filler word and pause marked
For a podcast episode you're turning into a blog post, manual transcription is almost never worth the time anymore.
What you get: Highest accuracy and control.
Cost: Your time, or $1-3 per minute if you hire a transcriptionist.
Time: 4-6 hours for a 45-minute episode.
Getting a usable transcript from AI output
Raw AI transcripts need editing before you publish or quote from them. The common issues:
No speaker labels. AI tools split by pause, not by speaker. You'll need to add "Host:" and "Guest:" at the start of each turn manually, or use a tool with speaker diarization like Otter.ai.
Proper nouns wrong. Names of people, products, and places often get substituted for phonetically similar words. "GPT-4" becomes "GBT for" and "Brené Brown" becomes "Renee Brown."
Filler words included. Depending on your purpose, you may want to remove "um," "uh," and "you know" before publishing.
A 45-minute interview transcript takes about 20-30 minutes to review and clean up — still much faster than manual transcription from scratch.
Which method to use
- Fastest: Check if a transcript already exists. If yes, you're done.
- Best accuracy / most common: Upload the MP3 to TranscribTxt or Whisper-based tool.
- No cost, lower quality: YouTube captions if the episode is on YouTube.
- Highest accuracy: Manual transcription or professional human transcription service.
Frequently Asked Questions
How do I get a transcript of a podcast episode I didn't record?
Download the podcast's MP3 file (most podcast apps let you download episodes), then upload it to a transcription tool like TranscribTxt. The whole process takes about 5 minutes plus the transcription time (roughly 1 minute per 10 minutes of audio). Some podcasts also publish transcripts on their website or on podscribe.com — worth checking before transcribing yourself.
How accurate is AI podcast transcription?
On a well-recorded podcast with one or two speakers, modern AI transcription (Whisper-based) is 94-96% accurate. That means roughly 1 error per 20 words. On podcasts with multiple overlapping speakers, music under speech, or heavy accents, accuracy drops to 85-90%. You'll need to review the transcript before publishing or using it.
Is there a free way to transcribe a podcast?
Yes. TranscribTxt's free plan gives you 5 files per month at no cost — enough to transcribe a few episodes. YouTube auto-captions are free if the podcast is on YouTube, though accuracy is lower. OpenAI Whisper is free to run locally but requires technical setup. Otter.ai offers 300 free minutes per month.
How long does it take to transcribe a 1-hour podcast?
With AI tools, a 1-hour podcast transcribes in 4-6 minutes. Manual transcription at professional speed takes 4-6 hours for a 1-hour recording. AI gets you a rough draft in minutes; you spend 20-40 minutes reviewing and correcting it. Total time with AI is 30-50 minutes versus 5-6 hours manually.
Can I transcribe a podcast to improve its SEO?
Yes, and this is one of the best reasons to transcribe podcasts. Search engines can't listen to audio, but they index text. Publishing a transcript (or even a cleaned-up edited version) turns your podcast into a text page that ranks for queries. A 45-minute interview typically contains enough content to rank for dozens of long-tail keywords.