PODCAST RSS · TRANSCRIPT TAG · 2026
The podcast:transcript tag — transcripts in your RSS feed
Your feed is what every podcast app, directory crawler, and AI assistant reads — and right now most of them cannot see a single word of your audio. One tag per episode fixes that. Here is the exact markup, the format that every app honors, the Apple opt-in that most guides skip, and the honest answer about what the feed can and cannot do for Google.
Why the feed needs a transcript at all
Podcast apps do not transcribe your audio. They index what the RSS feed gives them: episode title, description, and — if you provide it — a linked transcript file. Apple Podcasts auto-generates its own transcripts when you do not supply one, but automatic transcripts carry no speaker names, mangle proper nouns, and are Apple's, not yours. Every other platform is more conservative: most show nothing at all beyond the metadata.
The podcast:transcript tag is the Podcasting 2.0 answer: a
per-episode link to a hosted transcript file that apps can display, index, and search inside.
It is one of the best-supported tags in the entire podcast namespace — Apple built its
publisher-transcript feature directly on it, and major hosts (Buzzsprout, Captivate, Fireside,
RSS.com) accept it natively. Supplying your own file is also the only way to get
speaker names displayed in Apple Podcasts.
The SEO payoff compounds: episodes with full transcripts rank for materially more keywords than episodes with show notes alone, because the transcript exposes every sentence of the episode to indexing. The tag is how that transcript reaches the app layer; the episode page is how it reaches Google (more on that below).
The tag, copy-paste
Two parts: declare the Podcasting 2.0 namespace once on the
<rss> element, then add one transcript tag per episode.
<rss version="2.0"
xmlns:podcast="https://podcastindex.org/namespace/1.0">
<channel>
<item>
<title>Episode 42: The Future of Live Audio</title>
<enclosure url="https://example.com/episodes/42.mp3"
length="38212000" type="audio/mpeg"/>
<podcast:transcript
url="https://example.com/transcripts/42.vtt"
type="text/vtt" language="en"/>
</item>
</channel>
</rss>
The tag takes four attributes:
| Attribute | Required | What it does |
|---|---|---|
| url | Yes | Direct HTTPS link to the transcript file. Host it somewhere stable — your CDN or host; it must stay reachable for the life of the episode. |
| type | Yes | MIME type of the file: text/vtt, application/x-subrip (SRT), application/json, text/html, or text/plain. A missing or wrong type gets the tag ignored. |
| language | No | BCP-47 code for the transcript's language. Omit it and apps assume the feed's language. |
| rel | No | rel="captions" marks the file as closed captions (timed). Multiple transcript tags per episode are allowed — one per format or language. |
Multiple tags are the norm for a polished show: a VTT for apps, an SRT fallback, and an HTML transcript for clients that render rich text. If you ship only one, ship the VTT.
VTT, SRT, JSON, or HTML — and who honors what
Formats verified against the Podcasting 2.0 namespace spec, Apple's RSS guide, and 2026 app documentation.
| Format | MIME type | Speaker names | Timing | Best for |
|---|---|---|---|---|
| WebVTT (.vtt) | text/vtt | Yes | Yes | The default choice — every supporting app reads it, Apple prefers it, browsers caption web players with it natively. |
| SRT (.srt) | application/x-subrip | Yes | Yes | Fallback / reuse — same file doubles as YouTube subtitles. Apple accepts it when VTT is absent. |
| Podcast JSON | application/json | Yes (per segment) | Word-level | Apps with highlighted playback; carries per-segment speaker, start/end, and body fields. |
| HTML / plain text | text/html / text/plain | Markup only | No | Clients that just display text. No timing means no captions — pair it with a VTT. |
Apple Podcasts
Opt-in via Podcasts Connect. Prefers your VTT over SRT; replaces its auto-transcript with yours — speaker names included. Rejects files with non-speech cues or worse accuracy.
Pocket Casts, Castro, Podcast Addict
Pocket Casts renders VTT/SRT/JSON/HTML for all users. Castro supports all four formats (since Feb 2025) with no auto-generation — your file is the only transcript it has. Podcast Addict takes SRT/JSON/HTML.
Player FM, Fountain, Goodpods
All display RSS-linked transcripts — Player FM flags them with a transcript icon, Fountain supports on-demand premium transcripts. Directory and AI crawlers also read the feed.
The Apple opt-in (the step most guides skip)
Adding the tag is not enough for Apple. Publisher transcripts are opt-in per show: in Apple Podcasts Connect, enable using your own transcripts, and Apple will start ingesting the files your feed links instead of generating its own. Ingestion takes longer than the auto-transcript, so expect a delay on new episodes.
Apple's acceptance rules are stricter than the spec, and they matter for your transcription workflow:
Speech only
The transcript file should contain only spoken words. Broadcast-style cues — "[DOOR SLAMS]", "[SPOOKY CHORDS]" — get those portions dropped, and too many gets your file rejected entirely in favor of Apple's auto-transcript.
Accuracy gate
Apple quietly rejects a file it judges less accurate than its own transcript. That means a proofread pass on proper nouns, brand names, and guest names is not optional — it is the difference between your transcript shipping and Apple's shipping instead.
The payoff justifies the discipline: your transcript shows speaker names (auto-transcripts never do), you control spelling and terminology, and the same clean file serves the feed, the episode page, and the web player.
What the feed tag does not do
Honest limits, because half the guides oversell this: Google does not rank your RSS
feed. The feed tag drives in-app search, transcript display, and the directory/AI
crawlers that read feeds — but when someone Googles your episode topic, the thing that ranks is
the episode web page, and that page needs the transcript as crawlable HTML
text plus PodcastEpisode structured data.
So the 2026 setup is two layers from one transcript: publish the full transcript on the episode
page with schema for Google and AI answer engines, and link the same file from the feed with
podcast:transcript for the apps. Produce it once — locally,
on your own machine — and every surface gets it. Our guides to
local podcast transcription
and PodcastEpisode schema
cover the other layer in detail.
The workflow, end to end
- Transcribe locally. Record the episode, drop the file into VocalFuse, and get a Whisper-class transcript with speaker turns on your own GPU/CPU — nothing uploads, no per-minute meter, no account. Export SRT, VTT, and TXT from the same pass (free tier; Basic $5/mo dictation, Pro $10/mo adds AI notes).
- Clean for Apple. Remove sound-effect and music cues, fix proper nouns and guest names, keep speaker labels. This is the proofread pass that decides whether Apple uses your file or its own.
- Host the file over HTTPS next to your episode assets — stable URL, correct MIME type, reachable for the life of the episode. A CDN is fine; a dead link is worse than no tag.
- Add the tag per episode (snippet above), keep episode GUIDs stable across edits, and validate the feed with a feed validator before publishing.
- Publish the page layer. Full transcript + PodcastEpisode JSON-LD on the episode page, submit to Google Search Console, and let the two layers compound.
Feed hygiene while you are in there: put your show's primary keyword in the channel
<title>, write a 200–400 word channel
<description>, include
atom:link rel="self" pointing at the canonical feed URL, and
keep every enclosure URL on HTTPS. When the transcripts are in, the natural next tag is
podcast:chapters — your SRT timestamps already contain the data
(our podcast chapters guide covers the tag and the SRT-to-chapters conversion).
Related reading
Local podcast transcription
Produce the transcript files free on Windows — no upload, no per-minute meter, SRT/VTT/TXT export.
The local workflowPodcast episode schema
Copy-paste PodcastEpisode JSON-LD with transcript, timeRequired, and AudioObject — the page layer that ranks in Google.
Episode schema templatePodcast show notes generator
Turn the transcript into summary, chapters, and quotes — free tier vs Castmagic $29+/mo.
Show notes workflowPodcast transcription software compared
Eleven 2026 tools by verified price, speaker labels, and where your audio uploads — Descript, Otter, Castmagic, Sonix, Rev, TurboScribe vs local.
The 2026 comparisonDescript podcast transcription
What Descript's 2026 plans include for podcast transcription — and where the meter bites.
Descript plans decodedPodcast chapters
The next tag after the transcript: podcast:chapters JSON, Apple's rules, and a free SRT-to-chapters workflow.
Chapter markers guideFree voice transcription
The genuinely-free transcription options on Windows — unlimited, no upload, no trial clock.
Free transcription optionsExplore related AI note taking guides
Podcast RSS transcript tag — FAQ
What is the podcast:transcript tag?
A Podcasting 2.0 RSS extension tag that links each episode to a hosted transcript file. It lives inside the episode <item> block with a url and a type attribute (plus optional language and rel="captions"), and your feed declares the podcast namespace at the top. Podcast apps that support it — Apple Podcasts, Pocket Casts, Castro, Podcast Addict, Player FM, Fountain, Goodpods — then show or index the transcript instead of relying on their own auto-generation.
Which transcript format should I use — VTT or SRT?
VTT. It is the format every Podcasting 2.0 app supports, Apple prefers it over SRT when both are present, and browsers render VTT natively so the same file can caption a web player. Use SRT only as a fallback or when a downstream tool (like YouTube subtitles) demands it. The JSON transcript format adds word-level timing for highlighted playback; plain text/html transcripts work for apps that just display text but lose the timing.
How do I get Apple Podcasts to show my transcript instead of its auto-generated one?
Opt in inside Apple Podcasts Connect (per show), then link a VTT (or SRT fallback) file from each episode with the podcast:transcript tag. Apple matches your file to the audio and displays it with speaker names — something auto-transcripts do not show. Two rules: the file should contain only speech (sound-effect or music cues get Apple to quietly reject the file and fall back to their own), and Apple rejects files it judges less accurate than its own transcript, so proofread names and jargon before publishing.
What MIME type goes in the type attribute?
text/vtt for WebVTT, application/x-subrip for SRT (that is the MIME the Podcasting 2.0 spec lists — not application/srt, though Apple also accepts it), application/json for the PodcastIndex JSON transcript schema, and text/html or text/plain for HTML/text transcripts. A valid type attribute is required — feeds with a missing or wrong type get the tag ignored.
Does a transcript in the RSS feed help Google rankings?
Not directly. Google does not rank the RSS feed — it ranks the episode web page, and that page needs the full transcript as crawlable HTML text plus PodcastEpisode structured data. The feed tag does three different jobs: it drives in-app search and transcript display in podcast apps, it feeds AI assistants and directory crawlers that read feeds, and it signals episode substance to platforms that generate their own (worse) transcripts. Do both: transcript on the episode page for Google, podcast:transcript in the feed for the apps.
Which podcast apps support podcast:transcript?
Apple Podcasts (opt-in via Podcasts Connect, VTT preferred, overrides its auto-transcript), Pocket Casts (VTT, SRT, JSON, and HTML for all users), Castro (all four formats since February 2025, no auto-generation), Podcast Addict (SRT, JSON, HTML), Player FM (shows a transcript icon for RSS-linked transcripts), Fountain, and Goodpods. Major hosts — Buzzsprout, Captivate, Fireside, RSS.com — accept the tag, so the work is producing one clean VTT per episode.
Can I generate the transcript files without paying per minute?
Yes. Local transcription runs a Whisper-class model on your own machine — no upload, no per-minute meter, no subscription just to export. VocalFuse transcribes on Windows and exports SRT, VTT, and TXT from the same pass, so the file you clean for Apple is the file you link in the feed and publish on the episode page. Free tier; Basic $5/mo for dictation, Pro $10/mo adds AI notes. Self-hosted open-source Whisper produces the same formats if you prefer rolling it yourself.