YouTube Transcript Scraper — Subtitles, Captions & LLM-Ready Text

Extract the transcript, subtitles and captions of any YouTube video — no API key, no browser, no sign-in. Feed it a single video, a list of thousands, a playlist or a whole channel, and get back clean timestamped segments, LLM-ready plain text, readable paragraphs, and SRT / VTT subtitle files.

Built for reliability first: a multi-client engine keeps working when YouTube shifts, and a surgical residential fallback rescues the few videos that datacenter IPs can't reach — so you get the uptime of premium tools while staying on the free datacenter path for almost everything.

What it does

This actor is a fast, resilient YouTube transcript API / subtitle extractor / caption downloader. Give it YouTube URLs (or bare video IDs, youtu.be links, Shorts, /live/, /embed/, playlist URLs, or channel URLs like /@handle, /channel/UC…, /c/name, /user/name) and it returns each video's transcript in the formats you choose. It handles manual and auto-generated (ASR) captions, lets you pick a language or translate, and pulls rich video metadata alongside the text.

Common searches this solves: youtube transcript, youtube subtitles download, youtube captions to text, youtube transcript api, bulk youtube transcript, youtube to srt, transcript for LLM / RAG / summarization.

Key features

• Resilient multi-client engine. Requests cascade across several YouTube clients (Android, iOS, mobile web, TV, web). When one surface changes, the others keep delivering — far sturdier than single-path scrapers that break on every YouTube update.
• Smart residential fallback. Most videos are fetched over the free datacenter proxy. Only the ones that trip YouTube's "confirm you're not a bot" check are retried over residential proxies, so you get near-100% success while paying the residential rate on just a handful of videos.
• Parallel processing. A configurable concurrency pool fetches many videos at once, each on its own rotating session — large playlists and channels finish quickly.
• Five output shapes. Timestamped segments, flat text (LLM-ready), grouped paragraphs (merged into readable prose, not choppy one-second lines), plus srt and vtt subtitle files.
• Rich metadata. Title, channel, channel ID, duration, view count, publish date, category, keywords, description, thumbnail, and the list of available caption languages.
• Language & translation. Ask for a specific language (with en matching en-US), fall back gracefully, or translate captions to another language.
• Playlist & channel expansion. Point it at a playlist or channel and it collects the videos for you, up to a cap you set.
• Never fails silently. Every input yields a record with an explicit status (found / not_found / error) and a precise error code — no mystery empty rows.

Input

Field	Type	Description
videos	array (required)	Video URLs, video IDs, youtu.be links, Shorts, /live/, /embed/, playlist URLs, or channel URLs. Playlists and channels are expanded into their videos.
language	string	Preferred caption language code (e.g. en, fr, es). Matches en-US when you ask for en. Empty = first available. Default en.
includeAutoGenerated	boolean	Allow automatic (ASR) captions when no manual captions exist. Manual is always preferred. Default true.
outputFormats	array	Any of segments, text, paragraphs, srt, vtt. Default ["segments", "text"].
translateTo	string	Translate captions to this language code using YouTube's own translation. Empty = keep original.
includeMetadata	boolean	Attach title, channel, duration, views, keywords, description, thumbnail, etc. Default true.
maxVideosPerChannel	integer	Cap on how many videos to take from each playlist/channel. Default 20.
maxConcurrency	integer	How many videos to fetch in parallel. Default 5.
proxyConfiguration	object	Proxy used to reach YouTube. Apify Proxy (datacenter) by default — free on the platform.
autoResidentialFallback	boolean	Retry bot-checked videos over residential proxies automatically. Default true.

Output

One dataset item per video. Example (found, with segments, text and paragraphs):

{
  "input": "https://youtu.be/jNQXAC9IVRw",
  "videoId": "jNQXAC9IVRw",
  "url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "status": "found",
  "language": "en",
  "isAutoGenerated": false,
  "availableLanguages": ["en", "de"],
  "clientUsed": "ANDROID",
  "video": {
    "title": "Me at the zoo",
    "channel": "jawed",
    "channelId": "UC4QobU6STFB0P71PMvOGN5A",
    "durationSeconds": 19,
    "viewCount": 398413203,
    "isLive": false,
    "publishDate": "2005-04-23",
    "category": "People & Blogs",
    "keywords": ["me at the zoo", "jawed karim", "first youtube video"],
    "description": "The first video on YouTube...",
    "thumbnail": "https://i.ytimg.com/vi/jNQXAC9IVRw/hqdefault.jpg"
  },
  "segments": [
    { "start": 0.85, "dur": 3.2, "text": "All right, so here we are, in front of the elephants" }
  ],
  "text": "All right, so here we are, in front of the elephants, the cool thing about these guys...",
  "paragraphs": [
    { "start": 0.85, "dur": 18.1, "text": "All right, so here we are, in front of the elephants..." }
  ],
  "error": null
}

Videos without captions return status: "not_found"; unavailable/blocked videos return status: "error" with a code such as NoCaptionsAvailable, VideoUnavailable, AgeRestricted, or IpBlocked.

Use cases

• Feed LLMs and RAG pipelines — turn any video or channel into clean text for summarization, Q&A, or embeddings.
• Repurpose content — pull transcripts to write articles, show notes, chapters, or social posts.
• Subtitle files — export ready-to-use SRT/VTT for editing or re-uploading.
• Research & analysis — batch-collect transcripts across a channel or topic for search and NLP.

How it works & reliability

YouTube no longer serves caption data to the plain web player without a proof-of-origin token, which is why many transcript tools broke in late 2025. This actor instead asks YouTube through its mobile-app and TV clients, whose caption URLs remain accessible, and rotates between them automatically. If an IP is challenged with a bot-check, the video is transparently retried over residential proxies. The result is high success rates without forcing residential costs on every request. When YouTube does change something, fixing it is usually a small update to the client table — and a daily healthcheck flags breakages early.

Limitations & fair use

• Only videos that actually have captions (manual or auto) can produce a transcript.
• Private, deleted, or membership-only videos can't be accessed; they're reported clearly.
• Auto-generated captions reflect YouTube's speech recognition and may contain errors.
• Please use transcripts in line with YouTube's Terms of Service and applicable copyright. You're responsible for how you use the extracted text.

Pricing

This actor is free — the developer charges nothing. You only pay for your own Apify platform usage (compute units and, when the residential fallback is triggered, a small amount of residential proxy traffic). Typical costs are a fraction of a cent per video.

More actors by this developer

Explore the rest of the collection on the author's Apify profile: https://apify.com/fit_melon

---

Feedback and feature requests are welcome via the Issues tab — response time is public and taken seriously.