YouTube Transcript Scraper: Bulk, Channels, Shorts
Pricing
from $3.20 / 1,000 transcripts
YouTube Transcript Scraper: Bulk, Channels, Shorts
YouTube transcript API: bulk transcripts and captions for videos, Shorts, channels, playlists and searches. Timestamps, text, SRT/VTT, RAG-ready chunks, language fallback, translation. $3.20 per 1,000 videos processed.
Pricing
from $3.20 / 1,000 transcripts
Rating
0.0
(0)
Developer
John Wilkinson
Maintained by CommunityActor stats
0
Bookmarked
5
Total users
1
Monthly active users
7 days ago
Last modified
Categories
Features
Share
YouTube Transcript Scraper & Downloader
Download YouTube transcripts, captions and subtitles in bulk. Paste video, Shorts, playlist or channel links (or search queries) and get one row per video with the transcript as plain text, timestamped segments, SRT, WebVTT or RAG-ready chunks. $3.20 per 1,000 transcripts, charged once per video processed (a video with no captions or an error costs the same); metadata-only runs cost $1.20 per 1,000.
This YouTube transcript extractor works on single videos, YouTube Shorts, whole channels, playlists and YouTube search results. Pick caption languages, prefer human-made over auto-generated captions, or use YouTube's built-in caption translation. Output is ready for RAG, LLMs, AI agents, n8n, Make, Zapier, LangChain and MCP, content repurposing and research: time-aligned chunks with &t=123s deep links for citations.
Features
- Bulk input: paste any mix of video URLs, Shorts URLs,
youtu.belinks, video IDs, channel URLs or@handles, playlist URLs, and search queries. - Whole channels and playlists: takes the newest N videos, Shorts, or live streams of a channel, with an optional published-after date filter.
- Shorts transcripts: Shorts work like regular videos, one row each.
- Timestamps: segments with
startanddurationin seconds, text with[mm:ss]markers, SRT, and VTT. - RAG-ready chunks: time-aligned
chunks[]withstart,end,text, and a&t=123sdeep link for citations. - Language control: a preferred-language list (
en,es,de…), human-made captions preferred over auto-generated, fallback to the original language, and optional translation (translateTo). - Reliable on Apify: residential proxy by default, a fresh IP on every retry, and only a few KB transferred per video, so it stays fast and cheap.
- Transparent pricing: one charge per video processed, and a clear
statusanderrorCodeon every row (no_captions,unavailable,private,age_restricted,blocked…). - Metadata included: title, channel, channel ID, duration, views, description, thumbnail, Shorts flag, available caption languages, and a metadata-only mode.
Input
| Field | Default | Description |
|---|---|---|
urls | – (the form prefills two sample videos) | Videos, Shorts, youtu.be links, IDs, channels (@handle, /channel/UC…, or a tab link, see below), playlists |
searchQueries | – | Get transcripts of top search results |
maxVideosPerSource | 20 | Newest N per channel tab, playlist, or search (max 1,000) |
publishedAfter | – | YYYY-MM-DD (approximate for listings) |
languages | ["en"] | Preferred caption languages, in order |
fallbackToAnyLanguage | true | Otherwise use the original language |
translateTo | – | e.g. en: YouTube auto-translation |
outputFormats | ["text","segments"] | Any of text, segments, timestampedText ([m:ss] every ~30 s), chunks (RAG), srt, vtt |
chunkSizeChars | 2000 | Size of RAG chunks |
metadataOnly | false | Skip transcripts and return video metadata only |
proxyConfiguration | Residential | Keep the default for best reliability |
Channel tabs: paste the link of the tab you want
The tab is read from the link, so there is no separate setting:
| You paste | You get |
|---|---|
https://www.youtube.com/@name, @name, UC… id, /channel/UC…, /c/…, /user/… | the channel's Videos tab (the default) |
https://www.youtube.com/@name/videos | Videos |
https://www.youtube.com/@name/shorts | Shorts |
https://www.youtube.com/@name/streams or /@name/live | Live streams (past and current) |
/@name/featured, /playlists, /community, /releases, /podcasts | treated like the plain channel link: Videos |
The same works for /channel/UC…/shorts, /c/…/shorts and /user/…/shorts; m.youtube.com links and ?si= share parameters are fine. To get several tabs of one channel, add one link per tab; they are listed together as one listing (charged once), with maxVideosPerSource applied to each tab. A link to a single Short (/shorts/<videoId>) or live video (/live/<videoId>) is a video, not a tab.
Example input
{"urls": ["https://www.youtube.com/watch?v=T-D1OfcDW1M","https://www.youtube.com/shorts/aNI2jv7lBPQ","https://www.youtube.com/@IBMTechnology","https://www.youtube.com/playlist?list=PLZHQObOWTQDPD3MizzM2xVFitgF8hE_ab"],"searchQueries": ["retrieval augmented generation explained"],"maxVideosPerSource": 10,"publishedAfter": "2025-01-01","languages": ["en"],"outputFormats": ["text", "chunks"]}
Output
One dataset item per video:
{"videoId": "T-D1OfcDW1M","url": "https://www.youtube.com/watch?v=T-D1OfcDW1M","status": "success","source": "https://www.youtube.com/watch?v=T-D1OfcDW1M","title": "What is Retrieval-Augmented Generation (RAG)?","channelName": "IBM Technology","channelId": "UCKWaEZ-_VweaEx1j62do_vQ","durationSec": 395,"isShort": false,"publishedDate": "2023-08-23","viewCount": 1500000,"language": "en","isAutoGenerated": false,"availableLanguages": ["en", "en (auto)"],"wordCount": 1062,"description": "Learn about RAG…","thumbnail": "https://i.ytimg.com/vi/T-D1OfcDW1M/hqdefault.jpg","text": "Large language models. They are everywhere...","segments": [{ "start": 0.0, "duration": 4.2, "text": "Large language models." }],"chunks": [{ "index": 0, "start": 0.0, "end": 98.4, "text": "...", "url": "https://www.youtube.com/watch?v=T-D1OfcDW1M&t=0s" }]}
statusissuccess,no_captions,error, ormetadata(metadata-only mode). Rows that did not succeed carry anerrorCode(no_captions,unavailable,private,age_restricted,blocked,byte_cap…) and a readableerror.publishedDate(YYYY-MM-DD) is exact for videos you request directly (one extra ~1 KB request, included in the price) and approximate for videos found through a channel, playlist or search (derived from YouTube's "3 weeks ago"). If the exact date cannot be fetched it isnull; the row is never failed or charged extra because of it.sourceshows which input a video came from.translatedToappears only whentranslateTois used.- A run that delivers no transcript at all because of operational failures (blocked by YouTube, timeouts, failed listings) or because every input was invalid or not found finishes as Failed, with the reason in the status message; the error rows are still in the dataset. Private, age-restricted, unavailable and no-caption videos are normal answers and do not fail a run.
- A row over Apify's 9 MB dataset item limit (extremely long videos with every output format on) is delivered without its bulkiest fields and carries
droppedFieldsandwarning. - Run statistics (
status, counts, elapsed time) are saved to theRUN_SUMMARYkey in the key-value store.
Pricing (pay per event)
| Event | Price |
|---|---|
| Actor start (per GB of run memory, minimum 1; runs use 256 MB) | $0.0023 |
| Transcript returned | $0.0032 ($3.20 per 1,000) |
| Video metadata (metadata-only mode) | $0.0012 |
| Channel, playlist or search listed | $0.0042 per source |
| Failed video or listing (no captions, unavailable, blocked, empty/failed listing) | $0.0032 |
| Failed video in metadata-only mode | $0.0012 |
You pay for work done, success or failure. Each video is charged once, when processing starts, so a video with no captions, an error or a block costs the same as a transcript. A channel, playlist or search is charged once when it finishes listing, or at the failed price if the listing fails. Entries that are not YouTube URLs, and duplicate videos, are skipped and not charged. A source is listed only while your maximum cost per run can cover it and the videos it returns. Set a maximum cost per run in Apify to cap spend; the run stops cleanly when it is reached. Batch many videos per run to keep the start fee small.
Use cases
- RAG / AI knowledge bases: index a channel's videos into a vector database with timestamped citations.
- Content repurposing: turn videos and Shorts into blog posts, newsletters, threads, and show notes.
- Research and monitoring: analyse what creators, competitors, or experts say over time.
- Subtitles: export SRT or VTT in any language YouTube provides or can translate to.
- AI agents and MCP: call it from Claude, ChatGPT, LangChain, n8n, Make, or Zapier through the Apify API or MCP server.
Call it from code
curl -X POST "https://api.apify.com/v2/acts/jcwilk~youtube-transcripts/run-sync-get-dataset-items?token=$APIFY_TOKEN" \-H "Content-Type: application/json" \-d '{"urls":["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],"outputFormats":["text"]}'
FAQ
Can I call it from code or an AI agent? Yes: use run-sync-get-dataset-items (example above), the Apify client libraries, or the Apify MCP server. A single video comes back in a few seconds.
Does it work for Shorts? Yes. Paste a /shorts/ URL, or the channel's Shorts tab link (https://www.youtube.com/@name/shorts) to take its newest Shorts.
What if a video has no captions? You get status: no_captions. The video was charged when processing started ($0.0032), like any other video. This Actor reads YouTube's own captions (manual or auto-generated); it does not run speech-to-text.
Which languages? Every caption track YouTube has for the video, plus YouTube's translation into ~100 languages via translateTo.
Why a residential proxy? YouTube blocks many datacenter IPs. Each video needs only a few KB through the proxy, so the cost is included in the price.
Limitations
- Private, members-only and some age-restricted videos cannot be read.
- Live streams have captions only after they end.
- Dates for videos found through channels, playlists and searches are approximate ("2 weeks ago").
- Proxy traffic is capped at 200 KB per video (40 KB in metadata-only mode). That covers nearly all videos, including multi-hour podcasts; a video that exceeds it fails with
byte_capand is charged as a failed video. Listings stop at the number of videos you asked for (at most 1,000 per source). - This Actor reads YouTube's own captions (manual or auto-generated); it does not run speech-to-text.
This tool reads publicly available caption data. You are responsible for using it in line with YouTube's terms and applicable law.