Podcast transcription

Podcast transcription, with show notes written from it

Drop in the audio and get a clean, speaker-labeled, timestamped transcript back in minutes, plus an AI summary you can paste as your episode description. 95%+ accuracy on clear audio, honest about the rest. The first 10 minutes of any file are free.

Transcribe your podcastFirst 10 minutes free. No credit card.

Hear it, read it, at the same time

This is a real transcript synced to its audio. Press play, or click any word to jump straight to that moment. Same thing happens to your episode.

Sample preview
play it · click any word to jump

Host:All right, welcome back to the show. So today, we are getting into something I think every single creator struggles with, which is, honestly, how do you actually finish the stuff you start?

Guest:Oh, this is so my topic because I am genuinely terrible at it.

Host:Right? So, okay, the thing that flipped it for me was time boxing. You give a task like 30 minutes, and when that timer goes off, you stop. Doesn't matter if it is done.

Guest:Wait, wait. Even if it is not finished?

Host:Especially if it is not finished. That is kind of the whole point. The deadline does the work.

Guest:Huh. Okay, that is actually kind of brilliant. I always thought I had to, you know, see the whole thing through in one sitting.

Host:Yeah. That is the trap. Little blocks every day beats one big heroic marathon that, let's be real, never actually happens.

0:00
1:01

Not just a transcript. The show-page material, ready to paste.

Every line is timestamped and speaker-labeled, so the transcript is ready for your show page and it's easy to read chapter markers off the topic changes. The AI summary gives you a clean prose recap for the episode description. A full transcript is also the biggest SEO win a podcast can get: an hour of audio Google can't read becomes an indexable page.

What you get backtranscript + summary
Timestamped transcript
  • 00:00 · HostWelcome back. Today we're getting into why file-first tools beat the live-meeting ones for podcasters.
  • 04:12 · GuestRight, because you record however you want, then hand off the finished file. The preview shows you the transcript before you pay a cent.
  • 11:38 · HostAnd the accuracy holds up on clear audio. They're honest about where it drops, which I respect.

Every line carries its moment, so chapter markers are just the topic changes. Speaker labels rename in one edit.

AI summary

A conversation on transcribing podcasts from recorded files rather than live tools. The hosts cover the preview-before-you-pay model, where you see the full transcript quality before paying, and discuss realistic accuracy: strong on clear audio, lower on crosstalk and compressed remote calls. They close on pricing and which plan fits a weekly show.

A clean prose recap, ready to paste as your episode description.

Answer the three things every podcaster asks first

How accurate will it be on my setup, how do I get my file in, and what does it cost. Straight answers, right here on the page.

What accuracy should you actually expect?

95%+ on clear audio. Where your setup lands depends on how it was recorded. Pick yours, honestly. See the live benchmark.

Two hosts, local-recorded (Riverside, Zencastr): Distinct voices on their own tracks stay close to the top. You review the flagged words in the editor, so the version you ship is right either way.

What will your show cost?

Drag to your real episode length. The price is the price.

45 min
4

Pay per episode

$3

pay once, $12/mo at this cadence

Basic plan

$10/mo

covers ~10 hrs of audio a month

At 4 episodes a month, the $10 Basic plan is cheaper than paying per episode for the same product (every plan has every feature).

Hours round up: a 1 hour 2 minute episode bills as 2 hours.

Get your episode out of any recording tool

Where did you record it?

  1. 1Open the recording in your Riverside dashboard.
  2. 2Download the audio: the mixed track is one file, or grab each host's track separately for the cleanest speaker split.
  3. 3Upload the file here. Separate tracks? Upload the mix; we label speakers automatically.

Tip: Riverside records each person locally, so its audio is about as clean as remote gets.

Podcast-specific things worth knowing

The stuff that actually decides whether your transcript is good, none of which is about the transcription tool.

Multitrack beats a mixed file for speaker labels

If you record on Riverside or Zencastr, each host lands on their own track. That is close to the best case for telling voices apart. A single mixed-down MP3 still works, but when two similar voices are glued into one channel, expect to fix a few speaker labels by hand. Either way, renaming "Speaker 1" to a real name updates it everywhere in one edit.

Intro music is fine; loud beds under speech are the seam

A music-only cold open just comes back as silence in the text, which is correct. Speech over a light music bed transcribes cleanly. It is only dense music mixed loud under talking that shades a few words, and the per-word confidence view flags exactly those so you are not re-reading the whole thing.

A remote guest on a phone costs a few points, and that's normal

A guest dialing in over a compressed call will transcribe a little worse than your studio mic. That is the audio, not the model. If it matters, ask guests to record their side locally (a voice memo works) and send it. Our interview transcription guide goes deeper on multi-person audio.

Publish the cleaned-up version, keep the verbatim one

For a show page, use filler-word cleanup so the transcript reads like writing instead of a court record. The original stays available if you ever need exactly what was said.

Podcast transcription FAQ

Updated July 2026
Upload the episode audio (MP3, WAV, M4A, or the video file) and FileToText returns a speaker-labeled, timestamped transcript in minutes, plus an AI summary you can paste as show notes. There is no software to install and no conversion step. The first 10 minutes of any file transcribe free on a free account; after that it is $3 per audio hour, billed per started hour, so a 45-minute episode costs about $3. Accuracy is 95%+ on clear audio and dips on loud music beds or heavy crosstalk, where the per-word confidence view flags the words to check. If you record multitrack on Riverside or Zencastr, speaker labels are close to perfect. Export the finished transcript as TXT, DOCX, SRT, or WebVTT, or keep a read-only share link. See the live monthly benchmark →
Yes, and you should. Export as TXT or DOCX and paste it into your show page, or drop the SRT alongside the episode. A full transcript is the single biggest SEO win for a podcast: it turns an hour of audio Google can't read into an indexable page, and it makes the episode accessible to deaf and hard-of-hearing listeners.
Every transcript is timestamped and synced to the audio, so each line carries the exact moment it was said. Reading chapter markers off the topic changes takes a minute, and click any word to jump playback there, which makes pulling clips for social trivial. The AI summary gives you a clean prose recap for the episode description.
It labels speakers automatically and a clean multi-host recording is close to the best case: distinct voices, mostly taking turns. Rename "Speaker 1" to a real name once and it updates everywhere. If several voices sound alike, set the speaker count at upload to help. Heavy talking-over-each-other is where every tool, ours included, needs a human pass, so a quick review fixes those lines.
Speech over a light music bed transcribes fine. A loud music-only intro just comes back as no text for those seconds, and dense music under talking can shade a few words. None of it breaks the transcript; you'd tidy the seams in the editor if you're publishing it word-for-word.
Cleaned up. The filler-word cleanup drops the ums and false starts so the published transcript reads like writing, not a court record. The original stays available if you ever need exactly what was said. For a searchable show page, cleaned is what you want.
Pay-as-you-go is $3 per audio hour, billed per started hour, so a 45-minute episode is $3 and a 90-minute one is $6, paid once per file with nothing renewing. If you publish weekly, a $10/month Basic plan covers about 10 hours of audio, which is more than most shows record. Use the calculator above with your real episode length. Not satisfied? Full refund, no questions asked.

Start transcribing here