Podcast Transcripts from RSS (Published, No Speech-to-Text)

Get podcast transcripts as text straight from podcast RSS feeds. Give the Actor podcast RSS feed URLs; it finds every episode that publishes a Podcasting 2.0 <podcast:transcript> file (SRT, WebVTT, JSON, HTML or plain text), downloads that exact file and returns clean plain text plus timestamped speaker segments together with episode metadata.

No audio download and no speech-to-text, so it is fast, cheap and returns the creator's own transcript word for word. It is built for AI agents and RAG pipelines as much as for people: one call, one row per episode, predictable JSON.

What you get per episode

Field	Meaning
podcastTitle, feedUrl	Podcast and the feed it came from
episodeTitle, guid, episodeUrl, publishedAt, audioUrl, durationSeconds, episodeNumber, season	Episode metadata
hasTranscript	true when a published transcript was downloaded and parsed
transcriptUrl, transcriptType, transcriptFormat, language	Which transcript file was used (vtt, srt, json, text, html)
text	Full transcript as plain text; one paragraph per speaker turn, Name: labels optional
segments	[{start, end, speaker?, text}] in seconds when the transcript has timing
segmentCount, wordCount, speakers	Quick stats
availableTranscripts	Every transcript alternative the feed lists
transcriptError	Why a metadata-only row has no transcript

When a feed offers several formats, the Actor picks the richest: WebVTT, then SRT, then JSON (word-level JSON is merged into readable segments), then plain text, then HTML.

Example row (shortened):

{
  "podcastTitle": "Buzzcast",
  "episodeTitle": "Your Podcast Idea Doesn’t Have To Be Perfect To Get Started",
  "publishedAt": "2026-10-30T12:00:00.000Z",
  "durationSeconds": 751,
  "hasTranscript": true,
  "transcriptFormat": "vtt",
  "language": "en-us",
  "text": "Jordan: Alban, do you remember when we did that Buzzcast episode...\n\nAlban: No. I have no recollection of it...",
  "segments": [{ "start": 0.08, "end": 8.08, "speaker": "Jordan", "text": "Alban, do you remember when we did that Buzzcast episode..." }],
  "wordCount": 2478,
  "speakers": ["Jordan", "Alban", "Kevin"]
}

How to use it

1. Paste one or more podcast RSS feed URLs into feedUrls. (Tip: a podcast's RSS link is usually on its website or host page. Name search is not offered, because Apple's robots.txt disallows automated use of its search pages and this Actor only fetches robots-allowed paths.)
2. Set maxEpisodesPerFeed (newest first; default 5).
3. Run. Each row is one episode with its transcript.

Useful options: onlyWithTranscripts (default on; off adds free metadata-only rows), publishedSince, includeSegments, includeSpeakerLabels, maxResults (total cap).

Which podcasts have transcripts?

Only podcasts that publish a transcript in their feed. In a 59-feed sample taken on 2026-11-04 (top Apple Podcasts results for 20 popular topics), 10 feeds (17%) carried transcript tags, and 44 of the 286 newest episodes (15%) had one. Hosts that publish them include Omny Studio (iHeart, Bloomberg, Fox Sports and other networks), Buzzsprout (when the creator enables transcripts), Podcast Index shows and some Spotify for Creators (Anchor) feeds. Most Megaphone, Simplecast, Acast, Art19 and Libsyn shows in the sample did not publish one. If a podcast has no published transcript, this Actor does not invent one; use a speech-to-text Actor for those.

Pricing

Pay per event: $2.00 per 1,000 transcripts (transcript event, $0.002 each). Metadata-only rows (episodes without a transcript, returned only when onlyWithTranscripts is off) are free. Failed or blocked downloads are never charged.

For comparison, speech-to-text costs roughly $0.006-$0.05 per audio minute; a one-hour episode here costs a fraction of a cent.

Good citizenship and limits

• Public feeds only: no logins, no cookies, no paywalled or private feeds.
• Respects robots.txt for every feed and transcript host and its crawl-delay; disallowed URLs are skipped and listed in the run SUMMARY (for example changelog.com disallows all bots, so its feeds are not read).
• Plain HTTP, no browser and no proxy. At least 1 second between requests to the same host. Private and internal network addresses are blocked.
• Files above maxResponseMb (default 15 MB) are refused rather than partly read.
• With onlyWithTranscripts on, up to about 10x maxEpisodesPerFeed (at least 100) newest episodes are scanned per feed to find transcripts.
• Transcripts are the publisher's text; check the podcast's terms before redistributing them.

Run records

The key-value store holds SUMMARY (rows, transcripts, episodes scanned, per-feed stats, stop reason and errors) and STATE (resume point; a migrated or restarted run continues without duplicate rows or double charges).