Only 7,743,206 podcast episodes point to a transcript file through the open Podcasting 2.0 tag, about 4.84% of the 160,072,993 episodes discovered in public feeds (Podcast Industry Insights, Podcasting 2.0 and Podcast Index Statistics 2026, our calculation). The apps moved much faster than the feeds: Apple says it has transcribed more than 100 million episodes in 13 languages (Apple Podcasts for Creators, March 12, 2025). Chapter markers are rarer still, and in April 2024 only 2.20% of newly published episodes carried any embedded chapter data (Livewire Labs, Podcast Chapter Stats 2024). We aggregated data from Podcast Industry Insights, Blubrry, Livewire Labs, Apple, Spotify Research, the World Health Organization, the U.S. Department of Justice, W3C, Hugging Face and peer-reviewed language-learning studies. The hosting side of the same infrastructure is covered separately in our podcast hosting platform statistics.
TL;DR
- 7,743,206 episodes link a transcript through the podcast:transcript tag, 4.84% of discovered episodes (Podcast Industry Insights 2026, our calculation).
- Shows using transcripts rose from about 50,000 in April 2024 to nearly 80,000 in April 2025 (Blubrry, April 2025 Monthly Statistics).
- 2.20% of new episodes had embedded chapters in April 2024, or 3.73% counting description timestamps (Livewire Labs, Podcast Chapter Stats 2024).
- The transcript and chapters tags were part of Phase 1 of the podcast namespace, which closed on November 15, 2020 (Podcast Index, Podcast Namespace repository).
- Apple has transcribed more than 100 million episodes, including its entire back catalog, in 13 languages (Apple Podcasts for Creators, March 12, 2025).
- Apple now creates chapters automatically for shows that lack them, except trailers and shows under ten minutes (Podnews, Apple Podcasts automatic chapters FAQ 2026).
- Indexing full transcripts lifted podcast search nDCG from 0.239 to 0.336 in a benchmark test (Spotify Research, PODTILE, CIKM 2024).
- 430 million people need rehabilitation for disabling hearing loss (WHO, Deafness and Hearing Loss fact sheet, March 3, 2026).
- About 15% of US adults, or 37.5 million people, report some trouble hearing (NIDCD, Quick Statistics About Hearing 2024).
- A text alternative for prerecorded audio-only content is a Level A requirement, the lowest WCAG conformance tier (W3C, Understanding WCAG 2.1 SC 1.2.1).
- The ADA Title II web deadline for large US state and local governments moved to April 26, 2027 (U.S. DOJ, interim final rule, April 20, 2026).
- The best English average word error rate on the Open ASR Leaderboard is 5.63 (Hugging Face, Open ASR Leaderboard paper 2025).
1. Transcripts and Chapters Inside RSS Feeds: Still a Single-Digit Share
Fewer than 1 in 20 indexed episodes carries an open transcript file. The gap with the platforms is the real story: Apple alone says it has transcribed more than 100 million episodes (see section 3), roughly 13 times the 7.7 million episodes whose own feed links a transcript (100,000,000 / 7,743,206 = 12.9). A transcript generated inside one app serves only that app’s listeners. A transcript linked in the RSS feed travels to every app, search engine and archive that reads the feed, and the creator controls its accuracy.
Show-level counts look healthier than episode-level counts, and the back catalog is the likely reason. A show that switches the tag on today adds transcripts going forward, while its older episodes usually stay bare. The nightly tracker at Podcast Industry Insights and the monthly Blubrry statistics report both point the same way: steady growth from a small base. The Livewire Labs chapter study adds the per-episode view for chapters, which are even scarcer than transcripts.
| Metric | Value | Source |
|---|---|---|
| Episodes linking a transcript file through the podcast:transcript tag | 7,743,206 | Podcast Industry Insights, Podcasting 2.0 Statistics 2026 |
| Share of all discovered episodes that carry a transcript tag (7,743,206 / 160,072,993) | 4.84% | Calculated from Podcast Industry Insights, Podcasting 2.0 and Podcast Index Statistics 2026 |
| Podcasts using chapters / episodes with chapters | 65,185 podcasts / 1,615,598 episodes | Podcast Industry Insights, Podcasting 2.0 Statistics 2026 |
| Episodes offering transcripts, February 2025 snapshot | over 4.6 million | Blubrry, February 2025 Statistics |
| Shows using transcripts, April 2024 compared with April 2025 | about 50,000 vs nearly 80,000 | Blubrry, April 2025 Monthly Statistics |
| Shows using chapters, April 2024 compared with May 2025 | about 30,000 vs approximately 45,000 | Blubrry, April 2025 Monthly Statistics |
| New episodes with any embedded chapter data (ID3, namespace, Podlove or M4A format), April 2024 | 2.20% (about 1 in 45) | Livewire Labs, Podcast Chapter Stats 2024 |
| New episodes with chapters when timestamps written in the episode description are also counted, April 2024 | 3.73% (about 1 in 27) | Livewire Labs, Podcast Chapter Stats 2024 |
Context: chapters reach an even smaller share of the back catalog, 1,615,598 of 160,072,993 discovered episodes, or 1.01% (calculated from Podcast Industry Insights 2026). Most recent available data for the per-episode chapter share: Livewire Labs, April 2024, based on about 1.8 million new episodes published that month.
2. Podcasting 2.0 Namespace Adoption Beyond Transcripts
The podcast namespace is a community-run set of extra RSS tags maintained alongside the Podcast Index, an open podcast directory. Transcript and chapters were in its very first batch of tags, closed on November 15, 2020, so the low adoption in section 1 is not a story of a young standard. A tag only becomes official once at least 1 host and 1 app commit to it, or once it is already used in the wild (Podcast Index, Podcast Namespace repository). That rule guarantees a foothold, not scale.
The pattern in the counts is effort. The funding tag is a single link to a support page, and it is on 93,858 podcasts. Value blocks (tags that tell an app where to route listener micropayments, the so-called value-for-value model) need a wallet and payment setup, and they sit on about a third as many shows (33,970 / 93,858 = 36%). Chapters show a second friction: format fragmentation. In April 2024 the older ID3 format, which embeds chapters inside the MP3 file, was still about 1.5 times as common as the namespace’s own JSON chapter file (Livewire Labs 2024).
| Metric | Value | Source |
|---|---|---|
| Date Phase 1 of the podcast namespace closed (5 tags, including transcript and chapters) | November 15, 2020 | Podcast Index, Podcast Namespace repository |
| Date the most recent closed phase ended (Phase 7: publisher and chat tags) | July 1, 2024 | Podcast Index, Podcast Namespace repository |
| Minimum commitment for a tag to be adopted (when it is not already in use) | 1 host and 1 app | Podcast Index, Podcast Namespace repository |
| Podcasts using the funding tag | 93,858 | Podcast Industry Insights, Podcasting 2.0 Statistics 2026 |
| Podcasts publishing value blocks (listener micropayment routing) | 33,970 | Podcast Industry Insights, Podcasting 2.0 Statistics 2026 |
| Feeds implementing value blocks, February 2025 snapshot | 24,890 | Blubrry, February 2025 Statistics |
| Share of new episodes by chapter format, April 2024: ID3 / namespace JSON / Podlove | 1.93% / 1.26% / 0.39% | Livewire Labs, Podcast Chapter Stats 2024 |
Context: Phase 8 of the namespace is still open, with one tag (image) listed so far, so the tag list itself keeps growing while the 2020 tags still reach only a single-digit percentage of episodes.
3. Platform Auto-Transcripts and Auto-Chapters at Apple and Spotify
The platforms did not wait for feeds to catch up. In about a year, Apple Podcasts went from 4 transcript languages to 13 and from new episodes only to the full back catalog. More than 100 million episodes now have an Apple-generated transcript (Apple Podcasts for Creators, March 12, 2025). Spotify generates transcripts and chapters for every new episode from eligible creators, and Apple added automatic chapters with iOS 26.2.
What this changes is the job of the feed tag. When coverage is automatic, a creator-supplied transcript or chapter file is no longer about having text at all; it is about control: correct names, correct spelling of guests and brands, chapter titles that match the show’s structure. Apple says podcasters who supply their own transcript can choose to display it instead of the generated one, and both Apple and Spotify generate chapters only where the creator has not provided them. Podnews, a trade newsletter, wrote that after iOS 26.2 almost every podcast now has chapters in Apple Podcasts, which is why its automatic chapters FAQ expects chapter use to rise.
| Metric | Value | Source |
|---|---|---|
| Languages supported when Apple Podcasts transcripts launched with iOS 17.4, March 2024 | 4 (English, French, Spanish, German) | Apple Newsroom, March 2024 |
| Countries and regions where Apple transcripts were available at launch | over 170 | Apple Newsroom, March 2024 |
| Apple episodes with transcripts after 8 more languages were added, October 2024 | more than 80 million | Podnews, reporting Apple figures, October 3, 2024 |
| Apple episodes transcribed, including all previously published episodes, March 2025 | more than 100 million | Apple Podcasts for Creators, March 12, 2025 |
| Languages supported by Apple transcripts, March 2025 | 13 | Apple Podcasts for Creators, March 12, 2025 |
| Minimum show length before Apple generates automatic chapters (trailers excluded) | 10 minutes | Podnews, Apple Podcasts automatic chapters FAQ (updated April 21, 2026) |
| Spotify episodes that receive automatic transcripts and chapters | every new episode from eligible creators (chapters only where the creator has supplied none) | Spotify for Creators, Automated transcripts and chapters, September 9, 2025 |
Context: Apple’s 80 million figure was reported by Podnews on October 3, 2024, citing Apple; the 100 million figure comes from Apple’s own creator site. Neither platform publishes how many listeners open transcripts or tap chapters.
4. Search and Discovery: What Text Does for Findability
The cleanest measurement of discovery effects comes from Spotify’s own researchers. In the PODTILE paper (CIKM 2024), they tested podcast search on the public TREC podcast benchmark (a shared research test set of about 100,000 episodes and 54 queries) using a standard keyword search method. Adding full transcripts to the episode description raised nDCG, a ranking-quality score from 0 to 1, from 0.239 to 0.336: a 40.6% relative gain ((0.336 - 0.239) / 0.239). Adding only auto-generated chapter titles gave a smaller but statistically significant 15.5% gain ((0.276 - 0.239) / 0.239), with an index the authors describe as more than ten times smaller than the transcript index.
The second finding matters for smaller creators. The paper reports that auto-generated chapters drew relatively more engagement on less popular shows than on popular ones, which suggests that machine-made structure helps listeners sample shows they do not already know. Average episode transcripts in Spotify’s data run to about 16,000 tokens (word pieces), which is why a short chapter list is a cheaper navigation layer than full text.
| Metric | Value | Source |
|---|---|---|
| Podcast search ranking quality (nDCG), episode description only | 0.239 | Spotify Research, PODTILE (CIKM 2024) |
| nDCG, description plus auto-generated chapter titles | 0.276 (statistically significant gain) | Spotify Research, PODTILE (CIKM 2024) |
| nDCG, description plus full transcript (index more than 10 times larger than with chapter titles) | 0.336 | Spotify Research, PODTILE (CIKM 2024) |
| Change in chapter-initiated plays after Spotify’s limited auto-chapter roll-out (April 2024) | +88.12% | Spotify Research, PODTILE (CIKM 2024) |
| Engagement with auto-generated chapters relative to creator-written chapters: less popular shows vs more popular shows | about 0.75 vs about 0.53 | Spotify Research, PODTILE (CIKM 2024) |
| New episodes whose descriptions use the timestamp format that creates Spotify chapter blocks, April 2024 | 1.70% | Livewire Labs, Podcast Chapter Stats 2024 |
Context: the engagement comparison in PODTILE is observational, measured over 10 days with a 7-day moving average, not an A/B test. Most recent available data for the Spotify-format description share: Livewire Labs, April 2024.
5. The Accessibility Case: Hearing Loss and the Rules That Require Text
The audience that cannot use audio alone is large and growing. The WHO counts 430 million people who need rehabilitation for disabling hearing loss, over 5% of the world’s population, and projects nearly 2.5 billion people with some degree of hearing loss by 2050 (WHO, Deafness and hearing loss fact sheet, March 3, 2026). Hearing loss also rises sharply with age, from 5% disabling hearing loss among US adults aged 45 to 54 to 55% at 75 and older (NIDCD 2024). When Apple launched transcripts, it said deaf and hard-of-hearing users could read an episode without pressing play at all.
The legal side is clearer than the market side. Under WCAG, the Web Content Accessibility Guidelines, a text alternative for prerecorded audio-only content is required at Level A, the minimum tier (W3C, Understanding SC 1.2.1). The US ADA Title II rule adopts WCAG 2.1 Level AA for state and local governments, so a podcast that a public university or city publishes on its website falls under it. The DOJ pushed the deadlines back one year in an interim final rule published April 20, 2026, but did not change the standard itself. Captions follow the same logic in video, as our short-form video caption usage statistics show.
| Metric | Value | Source |
|---|---|---|
| People worldwide who need rehabilitation for disabling hearing loss | 430 million (over 5% of the world’s population) | WHO, Deafness and hearing loss fact sheet, March 3, 2026 |
| People projected to have some degree of hearing loss by 2050 | nearly 2.5 billion | WHO, Deafness and hearing loss fact sheet, March 3, 2026 |
| US adults aged 18 and over reporting some trouble hearing | about 15% (37.5 million) | NIDCD, Quick Statistics About Hearing (updated September 20, 2024) |
| US adults aged 75 and older with disabling hearing loss | 55% | NIDCD, Quick Statistics About Hearing (updated September 20, 2024) |
| WCAG 2.1 conformance level of SC 1.2.1 (text alternative for prerecorded audio-only content) | Level A | W3C, Understanding WCAG 2.1 SC 1.2.1 |
| ADA Title II web compliance date for state and local governments serving 50,000 people or more | April 26, 2027 (previously April 24, 2026) | U.S. DOJ, interim final rule, Federal Register, April 20, 2026 |
| ADA Title II compliance date for smaller governments and special district governments | April 26, 2028 (previously April 26, 2027) | U.S. DOJ, interim final rule, Federal Register, April 20, 2026 |
| Original 10-year cost estimate of the 2024 rule / present-value savings from the one-year delay (7% discount rate) | $24.7 billion / about $2.775 billion | U.S. DOJ, interim final rule, Federal Register, April 20, 2026 |
Context: the NIDCD also reports that fewer than 1 in 3 (30%) adults aged 70 and older who could benefit from hearing aids has ever used them, so many listeners with hearing loss rely on the content itself, not on devices, to follow audio.
6. Transcript Accuracy and Reading Along: What the Research Shows
Automatic transcripts are only as useful as their word error rate (WER), the share of words a speech recognizer gets wrong. On the Open ASR Leaderboard, a Hugging Face benchmark of 64 models from 18 organizations, the best English average WER is 5.63, roughly one wrong word in every 18 (100 / 5.63 = 17.8). Those scores come from standard test sets; the paper does not measure noisy multi-speaker podcast audio with crosstalk and music beds, so treat them as a ceiling for clean speech. For a sense of training scale, OpenAI’s Whisper learned from 680,000 hours of weakly labeled audio. The research side of automatic audio description is covered in our automated audio captioning benchmark statistics.
Do transcripts help people understand more? The best recent evidence comes from second-language learners. A within-participant study of 46 English learners found better comprehension when reading while listening than when listening only (Hui, The Modern Language Journal 2024). A preregistered follow-up with 86 learners found reading while listening scored below silent reading, though both beat listening alone (Hui and Godfroid, Language Learning 2025). The practical reading: a transcript beside the audio is a clear gain over audio alone, which is exactly the situation of a podcast listener.
| Metric | Value | Source |
|---|---|---|
| Hours of weakly supervised audio used to train OpenAI’s Whisper speech recognition model | 680,000 hours | Radford et al., Robust Speech Recognition via Large-Scale Weak Supervision, arXiv 2022 |
| Models on the Open ASR Leaderboard as of October 8, 2025 | 64 models from 18 organizations | Hugging Face, Open ASR Leaderboard paper, arXiv 2025 |
| Best English average word error rate (NVIDIA Canary Qwen 2.5B) | 5.63 | Hugging Face, Open ASR Leaderboard paper, arXiv 2025 |
| Lowest average word error rate for Spanish / German | 3.22 / 4.50 | Hugging Face, Open ASR Leaderboard paper, arXiv 2025 |
| Comprehension when reading while listening compared with listening only (46 English learners) | better with reading while listening; no main effect vs reading only | Hui, The Modern Language Journal 2024 |
| Comprehension when reading while listening compared with silent reading (86 Chinese learners of English, registered report) | lower with reading while listening; both conditions beat listening only | Hui and Godfroid, Language Learning 2025 |
Context: the Whisper paper is from December 2022 and is listed here as the reference point for training scale, not as a current accuracy benchmark. Most recent available data for that row: Radford et al., 2022.
Summary: Podcast Transcript and Chapter Marker Adoption by the Numbers
| Metric | Value | Source |
|---|---|---|
| Episodes linking a transcript through the podcast:transcript tag | 7,743,206 (4.84% of discovered episodes) | Podcast Industry Insights 2026 (share calculated) |
| Podcasts using the funding tag | 93,858 | Podcast Industry Insights 2026 |
| Shows using transcripts, April 2025 | nearly 80,000 | Blubrry, April 2025 Monthly Statistics |
| New episodes with embedded chapter data, April 2024 | 2.20% | Livewire Labs, Podcast Chapter Stats 2024 |
| Date Phase 1 of the namespace (transcript and chapters tags) closed | November 15, 2020 | Podcast Index, Podcast Namespace repository |
| Apple episodes transcribed, March 2025 | more than 100 million | Apple Podcasts for Creators, March 12, 2025 |
| Languages supported by Apple transcripts, March 2025 | 13 | Apple Podcasts for Creators, March 12, 2025 |
| Minimum show length before Apple generates automatic chapters | 10 minutes | Podnews, Apple Podcasts automatic chapters FAQ 2026 |
| Spotify episodes that receive automatic transcripts and chapters | every new episode from eligible creators | Spotify for Creators, September 9, 2025 |
| Podcast search nDCG, description only vs description plus transcript | 0.239 vs 0.336 | Spotify Research, PODTILE (CIKM 2024) |
| Change in chapter-initiated plays after Spotify’s auto-chapter roll-out | +88.12% | Spotify Research, PODTILE (CIKM 2024) |
| People needing rehabilitation for disabling hearing loss | 430 million | WHO fact sheet, March 3, 2026 |
| People projected to have some hearing loss by 2050 | nearly 2.5 billion | WHO fact sheet, March 3, 2026 |
| US adults reporting some trouble hearing | about 15% (37.5 million) | NIDCD, Quick Statistics About Hearing 2024 |
| WCAG 2.1 conformance level requiring a text alternative for prerecorded audio | Level A | W3C, Understanding WCAG 2.1 SC 1.2.1 |
| ADA Title II deadline for governments serving 50,000 people or more | April 26, 2027 | U.S. DOJ, interim final rule, April 20, 2026 |
| Best English average word error rate on the Open ASR Leaderboard | 5.63 | Hugging Face, Open ASR Leaderboard paper 2025 |
| Training audio for OpenAI’s Whisper model | 680,000 hours | Radford et al., arXiv 2022 |
Methodology and Sources
- Podcast Industry Insights (Daniel J. Lewis), Podcasting 2.0 Statistics: nightly counts of podcasts and episodes using namespace tags (transcripts, chapters, funding, value blocks).
- Podcast Industry Insights, Podcast Index Statistics: 160,072,993 discovered episodes, used as the denominator for the transcript and chapter shares.
- Blubrry, April 2025 Monthly Statistics and Blubrry, February 2025 Statistics: show and episode counts for transcripts, chapters and value blocks.
- Livewire Labs, Podcast Chapter Stats: chapter frequency by format among new episodes, January 2022 to April 2024.
- Podcast Index, Podcast Namespace repository: tag phases, closing dates and adoption rules.
- Apple Newsroom, Apple introduces transcripts for Apple Podcasts and Apple Podcasts for Creators, 100 million episodes transcribed.
- Podnews, Apple expands transcripts to eight more languages and Podnews, Apple Podcasts automatic chapters FAQ.
- Spotify for Creators, Automated transcripts and chapters.
- Ghazimatin et al. (Spotify Research), PODTILE: Facilitating Podcast Episode Browsing with Auto-generated Chapters, CIKM 2024.
- WHO, Deafness and hearing loss fact sheet and NIDCD, Quick Statistics About Hearing.
- W3C, Understanding WCAG 2.1 Success Criterion 1.2.1 and U.S. DOJ, interim final rule extending ADA Title II compliance dates, Federal Register, April 20, 2026.
- Srivastav et al. (Hugging Face), Open ASR Leaderboard, arXiv 2025 and Radford et al., Robust Speech Recognition via Large-Scale Weak Supervision, arXiv 2022.
- Hui, Scaffolding comprehension with reading while listening, The Modern Language Journal 2024 and Hui and Godfroid, Listening, Reading, or Both?, Language Learning 2025.
- Data watch: the transcript and chapter shares divide two separate Podcast Industry Insights counts (namespace tag usage and discovered episodes), which are crawled independently and updated nightly, so treat 4.84% and 1.01% as approximate. Blubrry’s February 2025 snapshot counted 806,115 episodes with chapters, about half the 1,615,598 in the 2026 tracker; the two come from different crawls and dates, so we do not compute a growth rate from them. Blubrry’s show counts are rounded by the source (‘about’, ‘nearly’). The Livewire Labs per-episode data stops at April 2024, before Spotify and Apple switched on automatic chapters, so in-app chapter coverage today is far higher than any feed-level figure. Apple’s 80 million and 100 million transcript counts are company figures without independent audit, and neither Apple nor Spotify publishes how many listeners use transcripts or chapters. Open ASR Leaderboard error rates are measured on benchmark test sets, not podcast audio. The DOJ extension is an interim final rule and could be revised again.
Last updated: November 6, 2026. We update this roundup quarterly, and the next refresh is expected when Podcast Industry Insights posts its first 2027 Podcasting 2.0 tag counts and Apple publishes its next Apple Podcasts for Creators transcript milestone.