Podcast & Audio Transcriber

Turn podcast episodes and audio or video files into text with Whisper large-v3-turbo: a clean transcript, markdown with [hh:mm:ss] timestamps, SRT and WebVTT subtitles, and RAG-ready chunks. When a show already publishes its own transcript in its RSS feed, you get that instead, for a flat $0.002.

Paste RSS feeds or direct file links, and get one record per episode. Pay per minute of audio, and nothing for files that fail.

What people use it for

• RAG and AI agents: timestamped chunks ready to embed, so an answer can link to the exact minute of the episode.
• Show notes, SEO and newsletters: the full text of every new episode the day it comes out (monitoring mode).
• Subtitles: SRT or VTT files for video podcasts, clips and Shorts.
• Research and media monitoring: what a set of shows said about a company, product or topic, searchable.
• Accessibility: a text version of every episode for people who can't listen.

What it costs: a 30-minute episode costs $0.09 on a paid Apify plan, or $0.15 on the free plan. An episode whose feed already has a transcript costs $0.002. Failed downloads, files without speech and skipped episodes cost nothing.

Pricing

What you get	Paid Apify plans	Free plan
One started minute of audio transcribed with Whisper	$0.003	$0.005
A show's own transcript, read from its feed	$0.002 per episode	$0.002 per episode

You're charged after a transcript is ready. Billing uses the real length of the audio file (dynamically inserted ads count), per started minute, with a one-second grace. A run never spends more than the limit you set in Apify, and Max minutes to transcribe per run adds a cap of your own.

What you get

One record per episode or file. A real record (text shortened):

{
  "id": "d74a4412e8a0ff31",
  "source": "feed",
  "feedUrl": "https://rss.buzzsprout.com/2632213.rss",
  "podcastTitle": "On The Line with Jim Pyne",
  "episodeTitle": "Bruce Smith: 200 Sacks, 19 NFL Seasons & the Legacy of a Hall of Famer | On the Line with Jim Pyne",
  "audioUrl": "https://www.buzzsprout.com/2632213/episodes/19953675-bruce-smith-200-sacks-19-nfl-seasons-the-legacy-of-a-hall-of-famer-on-the-line-with-jim-pyne.mp3",
  "publishedAt": "2026-10-13T09:00:00Z",
  "transcriptSource": "publisher",
  "language": "en-us",
  "durationSeconds": 1668,
  "billedMinutes": 0,
  "transcriptFormat": "json",
  "wordCount": 4826,
  "text": "SPEAKER_00: Welcome to On the Line with Jim Pine. I'm Jim Pine. ...",
  "markdown": "[00:00:00] **SPEAKER_00** Welcome to On the Line with Jim Pine. ...",
  "chunks": [{"index": 0, "start": 0.16, "end": 68.16, "startTimestamp": "00:00:00", "speakers": ["SPEAKER_00", "SPEAKER_02"], "wordCount": 244, "text": "..."}],
  "srt": "1\n00:00:00,160 --> 00:00:02,000\nWelcome to On the Line with Jim Pine.\n\n2\n00:00:02,240 --> 00:00:03,040\nI'm Jim Pine.\n..."
}

transcriptSource is "publisher" when the show's own transcript was used and "whisper" when the audio was transcribed. Whisper records also carry durationSeconds, billedMinutes and model.

How it works

1. Reads each feed, newest episodes first, and applies your filters (date, title keywords, how many per feed).
2. Uses the show's own transcript when the episode links one (the Podcasting 2.0 podcast:transcript tag). It's exactly what the publisher wrote, and cheaper.
3. Otherwise transcribes the audio:
   • downloads the file and converts it to 16 kHz mono, which is what Whisper hears anyway;
   • cuts it into pieces of about 10 minutes, at pauses, so no word is split;
   • transcribes each piece with Whisper large-v3-turbo, passing on the end of the previous piece as context;
   • stitches the timestamps back together.
4. Builds the outputs: plain text, timestamped markdown, chunks, SRT/VTT and raw segments, whichever you choose.

Typical speed: a one-hour episode is ready in a few minutes.

What you can put in

Field	What it does
Podcast RSS feeds	Feed URLs of the shows. Most podcast directories and show websites list them.
Episodes per feed	How many of the newest matching episodes to take from each feed (default 3).
Audio or video file links	Public direct links to MP3, M4A, WAV, OGG, FLAC, MP4, WebM and anything else ffmpeg reads.
Published within (days) / Title keywords	Episode filters.
Use the show's own transcript	On by default. Turn off to always transcribe the audio.
Spoken language	en, de, es, pt-BR… Empty means auto-detect (about 100 languages).
Max minutes to transcribe per run	Your own cost cap. 0 means only the Apify spending limit applies.
Output options	Markdown, RAG chunks (and their size), SRT, VTT, raw segments.
Only new episodes + Monitor name	Monitoring mode (below).

Daily monitoring

Turn on Only new episodes and schedule the Actor daily. Each run returns only episodes it hasn't transcribed before, so you pay once per episode. Give each schedule its own Monitor name if you run several with different inputs.

Limits and fair use

• Up to 6 hours and 600 MB per file.
• On the free Apify plan, a run transcribes up to 60 minutes of audio. Shows' own transcripts don't count toward that, and paid plans have no such cap.
• No YouTube, TikTok, Instagram, Facebook, X, Vimeo, SoundCloud or Spotify links. Those are web pages, not audio files, and their terms don't allow downloading the media. Use RSS feeds or direct file links.
• Only transcribe audio you have the right to use. A transcript of someone's show is still their content: credit the show if you quote or republish it.
• No speaker names yet: Whisper doesn't tell speakers apart. Shows' own transcripts often do include speaker names, and those are kept.
• Accuracy is very good on clear speech. It drops with loud music, crosstalk or a lot of jargon. Setting Spoken language helps with short clips.

Privacy and data handling

• Audio is downloaded into the run's temporary storage, converted, and sent to Groq's speech-to-text API (Whisper large-v3-turbo) for transcription.
• The temporary files are deleted as soon as each episode is done.
• Transcripts are stored only in your own Apify dataset.

FAQ

Can I transcribe YouTube videos? No. See Limits: their terms don't allow it.

Which languages? Whisper's ~100 languages, detected automatically. The detected language is in the language field.

What if a feed has a transcript but I want Whisper's? Turn off Use the show's own transcript.

How do I get subtitles for a video file? Put the file link in Audio or video file links and keep SRT subtitles on. Turn on WebVTT if your player wants VTT.

Can AI agents use it? Yes, through the Apify API, the Apify MCP server, Make, Zapier or n8n. A record is plain JSON with the text, chunks and timestamps.

Related Actors

More tools from the same developer:

• Podcast Transcript Scraper: only the transcripts publishers already put in their feeds, $0.002 each, no audio processing.
• Visa Sponsorship Jobs API: Greenhouse, Lever, Ashby, SmartRecruiters and Recruitee boards with sentence-level visa tags.
• Workday Jobs Scraper: any Workday careers site, with visa tags backed by official H-1B and UK sponsor records.
• Remote Jobs by Country: four remote boards in one feed, filtered to jobs open to your country.
• UK Visa Sponsor Jobs: UK jobs from Reed's official API, checked against the Home Office register of licensed visa sponsors.