Short answer: type the podcast's name into Podcast Transcriber, an Apify Actor. It finds the show through Apple's podcast directory API, reads the show's RSS feed, and transcribes the newest episodes with OpenAI's Whisper model, which runs inside the Actor. Each row is about 30 seconds of speech with a start time. It costs $0.025 per minute of audio, with no API key and no audio upload.
What you need
| Input | Example | Notes |
|---|---|---|
| Podcast name | NASA Daily |
Every word you type must appear in the show's title, so a fuzzy match never transcribes the wrong show |
| or Apple Podcasts link | https://podcasts.apple.com/us/podcast/the-daily/id1200361736 |
The bare id 1200361736 also works |
| or RSS feed | https://feeds.simplecast.com/Sl5CSM3S |
Use this for shows that are not on Apple Podcasts |
How it works, step by step
- Name to feed. It queries Apple's iTunes Search API, which is public, keyless and documented for this purpose. The search is fuzzy: on 31 October 2026, "NASA Daily" ranked a different show, Nasa Kiwe Tekh Ksxaw, first. So the tool only accepts a result when every typed word appears in the title.
- Feed to episodes. It reads the show's own RSS feed, the same file every podcast app reads, and takes the newest episodes.
- Episode to text. It runs Whisper, which its authors trained on "680,000 hours of multilingual and multitask supervision" (Radford et al., 2022). The audio is streamed, and decoding stops at your minute limit, so a two-hour episode with a 10-minute limit is never downloaded in full.
Cost and speed (measured, 31 October 2026)
| This Actor | Other Apify transcription Actors | |
|---|---|---|
| Price per audio minute | $0.025 | $0.010 – $0.048 |
| API key needed | No | Varies |
| Speed on Apify (4 GB, 1 CPU core) | about 2× real time | — |
| Input you can type | a podcast name | usually a file URL |
Our compute cost on Apify was $0.0057–0.0068 per audio minute across two test runs, about 27% of the price.
Call it from code or from an AI agent
curl -X POST "https://api.apify.com/v2/acts/appdata-labs~podcast-transcriber/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
-H "Content-Type: application/json" \
-d '{"podcasts": ["The Daily"], "episodesPerPodcast": 1, "maxMinutesPerEpisode": 10}'
That is Apify's run-sync-get-dataset-items endpoint. It returns the rows as JSON when the run ends. AI agents that use the Apify MCP server can find and call it as a tool. Each row is under 4 KB, so it fits an agent's context.
Each row contains podcast, episodeTitle, publishedAt, startTime, startSeconds, endSeconds, text, passage, passageCount and status.
FAQ
Can it transcribe YouTube or Spotify-exclusive shows?
No. It reads public podcast feeds. Spotify-exclusive and video-only content has no public feed.
Which languages?
20, including English, Spanish, German, French, Japanese, Chinese, Arabic, Hindi and Thai. Set the language explicitly: Whisper does not detect it here.
How accurate is it?
It uses Whisper base. It is strong on clear speech, and weaker on crosstalk and music. It can invent words over jingles, and those passages come through as written.
What if the podcast is not found?
The run skips it, names the closest titles, and charges nothing.
I have audio or video files, not a podcast.
Use Audio & Video Transcriber, which works the same way for any MP3, M4A, WAV or MP4 URL.
Ready to run: The Daily, latest episode · any podcast by name · MP3 to text
Top comments (0)