Podcast Transcript & Episode Scraper - from RSS, no audio models

Get podcast episodes with the transcripts the publishers already made public: text from podcast:transcript files (SRT, VTT, JSON, HTML, plain text), transcript pages linked in the show notes, plus chapters, hosts and guests, sponsor reads, links and a keyword search across episodes. Give it RSS URLs, Apple Podcasts links or just show names.

Not speech-to-text. This Actor never listens to audio. If a show publishes no transcript, the episode row says transcriptStatus: "none" and explains it, instead of pretending. Coverage varies a lot by show - the podcast row tells you how many of the returned episodes had one.

What you get

A podcast row per show (title, author, hosts, categories, funding link, transcript coverage %) and an episode row per episode:

Group	Fields
Episode	title - publishedAt - episodeNumber - season - episodeType - duration / durationMinutes - audioUrl - link - explicit - image - guid
Transcript	transcriptStatus (published, published-via-show-notes-link, link-in-show-notes, tag-unreachable, available-not-fetched, none) - transcript (text) - transcriptFormat - transcriptUrl - transcriptLanguage - transcriptWordCount - transcriptSegments (optional, with start/end/speaker)
People	hosts - guests with guestsSource: podcast:person tags (reliable) or guessed from title/show notes (unverified)
Structure	chapters (podcast:chapters file, or timestamp lists found in show notes, with chaptersSource) - soundbites
Notes	showNotes (plain text) - links (URLs with link text) - sponsorMentions (lines that read like sponsor reads, with links)
Search	keywordMatches: per keyword the count, where it was found (title / showNotes / transcript) and snippets with the time in the episode

Which shows have transcripts?

Checked when this was written: Podcasting 2.0 (SRT for every episode, chapters and guests), Acquired (text/VTT on recent episodes), The Changelog (HTML transcript pages for part of its back catalogue). Many big publishers (NPR, NYT, Vox) publish none in their feeds - you will see none for those, with chapters/guests/links still extracted.

Example

{ "type": "episode", "podcastTitle": "Podcasting 2.0", "title": "Episode 273: \"Advertising is the Horse and Buggies of Podcasting\"",
  "publishedAt": "2026-10-02T00:00:00.000Z", "durationMinutes": 90, "hosts": ["Adam Curry", "Dave Jones"],
  "guests": [], "chaptersSource": "podcast:chapters", "chapters": [ { "startTime": 0, "title": "Intro" } ],
  "transcriptStatus": "published", "transcriptFormat": "srt", "transcriptWordCount": 13227,
  "transcript": "Adam Curry: ...", "keywordMatches": [ { "keyword": "podcast index", "count": 9, "foundIn": ["showNotes", "transcript"],
    "snippets": [ { "field": "transcript", "time": 460, "text": "…maybe the podcast index can benefit from that…" } ] } ] }

time is the start (in seconds) of the transcript paragraph containing the hit.

Input

Field	Default	Meaning
feeds	Podcasting 2.0 + Acquired	RSS URLs, Apple Podcasts links (.../id123456789) or show names (best Apple Podcasts match is used)
maxEpisodesPerFeed	5	Newest episodes returned per show
keywords	none	Keep only episodes mentioning a keyword in title, notes or transcript; adds keywordMatches
maxEpisodesToScan	50	With keywords: how many newest episodes per show are checked (their transcripts are downloaded)
since, minDurationMinutes	none	Date and length filters
onlyEpisodesWithTranscript	false	Skip episodes without a published transcript
fetchTranscripts / followTranscriptLinks	true / true	Download tagged transcripts; use transcript pages linked from show notes
includeTranscriptText, includeTranscriptSegments, maxTranscriptChars	true, false, 200000	Output size controls
fetchChapters, includeShowNotes	true, true	Extras

An empty input works: it returns the latest 5 episodes each of Podcasting 2.0 and Acquired, with full transcripts.

Use cases

• Search what shows said about a company, product or topic across many shows, with context and timestamps.
• Build a transcript corpus for research, RAG or fine-tuning from shows that publish transcripts (onlyEpisodesWithTranscript).
• Guest and sponsor research for PR, booking and ad intelligence (guests from podcast:person tags, sponsor lines with links).
• Content repurposing: chapters, links and notes per episode for newsletters and show-note tools.
• Feed for AI agents: clean JSON per episode; works with the Apify API, MCP and LangChain loaders.

Pipeline: podcast-transcript-scraper (keywords + transcripts) -> your vector DB, Sheets or an LLM summariser.

Pricing

Pay per event: $0.002 per show read (feed-scraped) and $0.002 per episode returned (episode-scraped), transcript included - there is no extra charge for transcripts. Ten episodes of one show cost $0.022; the default empty-input run (2 shows, 5 episodes each) costs $0.024. Shows that cannot be read are not charged.

FAQ

Can it transcribe episodes that have no transcript? No. It does not process audio. For those episodes you still get metadata, chapters, links, guests and the audio URL to send to a speech-to-text service of your choice.

What if a host blocks data-center IPs? The Actor retries with other user agents and then through Apify's datacenter proxy. Hosts that block both will show up as an error row for that show.

What about YouTube captions? Not supported: YouTube blocks server-side caption downloads.

How are guests found? Reliable when the feed has podcast:person tags. Otherwise a pattern search over the title and first lines of the show notes ("with", "feat.", "my guest is") fills guests and sets guestsSource to guessed (unverified).

How do I find a feed URL? Pass the Apple Podcasts link or the show name and the Actor looks the feed up.

Large feeds? Feeds of 20 MB are read as a stream and reading stops as soon as enough episodes are found.

Feedback: open an issue on the Actor page; a short review on the Store page helps other podcast researchers find it.

Related Actors

reviews-theme-analyzer - app-reviews-scraper - event-scraper.