Read the transcript of any podcast episode.
4.7 million podcasts to search, any episode of any of them. 7,092 episodes from the biggest shows are already transcribed and open instantly; anything else we transcribe and email you, usually within the hour. One credit is one hour of audio, and every account gets 10 a month, free.
4,730,997 podcasts searchable 7,092 episodes in the library 7,826 hours transcribed
- 01
Find the show
Every public podcast is searchable, not only the ones we index.
- 02
Unlock it, or request it
An episode we already have is 1 credit, any length, yours forever. One nobody has yet costs its length in hours. Every account gets 10 a month, free.
- 03
Get the link by email
Requests jump the queue. We email you the moment the transcript lands, one episode or twenty-five.
Top of the charts right now
Explore every genre- 1
The Daily - 2
Crime Junkie - 3
Dateline NBC - 4
Up First from NPR - 5
REAL AF with Andy Frisella - 6
Pardon My Take - 7
The Joe Rogan Experience - 8
Mick Unplugged - 9
The Dylan Gemelli Podcast - 10
Live Free with Josh Howerton - 11
Good Hang with Amy Poehler - 12
Morbid - 13
Pod Save America - 14
The Learning Leader Show With Ryan Hawk - 15
In The Dark - 16
Tomorrow, Today - 17
The Bill Simmons Podcast - 18
Coffeez with Joe Shalaby
Latest transcripts
new episodes, transcribed as they publish- Why the current tech backlash feels different
- The Fines
- From high school dropout to CEO, with Autodesk’s Andrew Anagnost
- If You Hate Selling, Watch This | Ep 998
- How the WNBA became a political prop in the culture wars
- #308: Zac Taylor, Head Coach, Cincinnati Bengals – Consistency creates results
- Best Peptides, Stem Cells, Testosterone & Exercise Hacks - Biohacking Roundtable - #1148
- Ryan Resch - How Data Changed the NBA and What AI Changes Next (Ep. 327)
One question sets every price: do we already have it?
One credit is one hour of audio. If we hold the transcript, the episode is 1 credit whatever its length: full text, timestamps, every export, forever. If we don't, transcribing it costs its length in hours, nearest hour, never less than one. AI Search is 1 a query on every plan. Nothing else moves a price.
Compare all plansBuilt for scripts as much as for readers
Everything on the site is one key and one balance away in the API: search, episode lists, transcripts as JSON, SRT or VTT with word-level timing, transcription on demand, and AI Search, which reads a whole episode and returns verbatim quotes with timestamps for whatever you ask. Same credits, same prices as the site.
Read the API docscurl -s https://podmenti.com/api/v1/ai-search/query \
-H "Authorization: Bearer pm_YOUR_KEY" \
-d '{"transcript_id": "a7c18c255b33d645",
"topic": "books recommended by title"}'
{ "found": true,
"mentions": [{
"quote": "Elizabeth Gilbert is the author of Eat Pray Love…",
"timestamp_seconds": 114 }, …] }Questions people ask first
- What does a transcript cost?
- One credit is one hour of audio. Every plan, Free included, is a monthly allowance of credits. An episode we already have is 1 credit whatever its length: the full transcript, timestamps and every export format. An episode nobody has transcribed yet costs its length in hours, rounded to the nearest hour and never less than one, because we have to actually transcribe it. Once an episode is yours it is 0 credits forever: open it again, download it again, never pay twice. AI Search is 1 a query. Finding shows and listing episodes costs nothing. The free plan's 10 credits are ten hours a month.
- How long does a new transcript take?
- Two speeds. Standard runs on our own machine: requests go to the front of its queue and a typical hour-long episode is ready within the hour, one credit per hour of audio. The fast lane runs the same episode on a rented GPU and is ready in a minute or two whatever its length, promised within 5 minutes for a typical job (big batches get a longer promise, stated up front), at 2 credits per hour. Miss the promise and the extra credits come back on their own. Either way we email you the link the moment it lands, up to 25 episodes at a time.
- Which podcasts can you transcribe?
- Any show with a public feed, which is nearly all of them: search covers the open Podcast Index, 4.7 million podcasts. The only ones we cannot do are app-exclusive shows that publish no public audio.
- What makes the transcripts?
- NVIDIA's Parakeet TDT 0.6B v3 speech model, running on our own hardware, with word-level timestamps straight from the recogniser. AI Search uses Google's Gemini Flash-Lite. Both are named and explained on the engine page.
- Are the timestamps accurate?
- Yes, and they stay accurate. Podcast audio changes after publication as ads are swapped in and out; we detect the change and re-align every timestamp automatically, on every plan.
- What formats can I export?
- Unlocking an episode covers every format: plain text, timestamped text, .srt, .vtt and word-level JSON, on the site and through the API, forever.