YouTube Transcript Scraper

Get the transcript (captions or subtitles) of any public YouTube video as plain text, timestamped segments or an SRT file. Paste video links, or a channel or playlist link to pull YouTube transcripts for many videos at once, ready for ChatGPT, Claude and other LLMs, RAG pipelines, summarizers and search indexes. No YouTube account, login or API key needed.

What you get

One dataset row per video with:

• Video metadata: video ID, URL, title, channel, channel ID, duration, view count, publish date
• Transcript: the full text, plus timestamped segments (start and duration in seconds) or an SRT subtitle string, depending on the output format
• Language info: transcript language code and name, original caption language, auto-generated flag, translated flag, and every caption track the video offers

Inputs and options:

• Videos, channels and playlists: watch URLs, youtu.be links, Shorts, live and embed URLs, or bare 11-character video IDs. Channel links (@handle, /channel/UC..., /c/..., /user/...) and playlist links expand into their videos, up to a limit you set.
• Preferred languages: a list such as en, de, es. en also matches en-GB and en-US. Human-made captions win over auto-generated ones. If none of your languages exist, you get the auto-generated track in the video's spoken language (or the first available track) instead of an empty result.
• Caption type: any, human-made only, or auto-generated only.
• Translation: give a target language and get YouTube's own machine translation of the captions.
• Output format: text plus segments, text only (smallest, best for LLMs), or text plus SRT.

Use cases

• LLM and RAG pipelines: feed video transcripts into summarizers, chatbots and vector databases.
• Content research: search and analyse what a channel has said across hundreds of videos.
• SEO and repurposing: turn videos into blog posts, show notes and social posts.
• Subtitles: download SRT files for editing or translation.
• Academic and media monitoring: build text datasets from lectures, talks and news channels.

How to use

1. Open the Actor in Apify Console and paste video, channel or playlist URLs.
2. Set Max videos per channel/playlist and your preferred languages.
3. Pick the caption type, output format and, if you want, a translation language.
4. Keep the residential proxy default and click Start.
5. Download the results as JSON, CSV, Excel or HTML, or read them through the Apify API.

Field	Description
urls	Video URLs or IDs, channel URLs, playlist URLs. Required.
maxVideos	Max videos taken from each channel or playlist (default 50). Channels are read newest first.
languages	Preferred language codes in order, default ["en"].
captionType	any (default, human-made preferred), manual, or auto.
translateTo	Optional language code for YouTube's machine translation. If a video doesn't offer it, you get the original language and isTranslated is false.
outputFormat	segments (default), text, or srt.
maxRetries	Retries per video on a fresh IP when blocked (default 6).
proxyConfiguration	Defaults to Apify residential proxy. YouTube blocks most datacenter proxies.

Input example

{
  "urls": [
    "https://www.youtube.com/watch?v=arj7oStGLkU",
    "https://www.youtube.com/@TED"
  ],
  "maxVideos": 20,
  "languages": ["en"],
  "outputFormat": "text"
}

Output example

{
  "videoId": "jNQXAC9IVRw",
  "url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "title": "Me at the zoo",
  "channel": "jawed",
  "channelId": "UC4QobU6STFB0P71PMvOGN5A",
  "durationSeconds": 19,
  "viewCount": 438283667,
  "publishDate": "2005-04-23T20:31:52-07:00",
  "language": "en",
  "languageName": "English",
  "originalLanguage": "en",
  "isAutoGenerated": false,
  "isTranslated": false,
  "availableLanguages": [{ "code": "en", "isAutoGenerated": false }, { "code": "de", "isAutoGenerated": false }],
  "text": "All right, so here we are, in front of the elephants the cool thing about these guys is that they have really... really really long trunks ...",
  "segments": [
    { "start": 1.2, "duration": 2.16, "text": "All right, so here we are, in front of the elephants" },
    { "start": 5.318, "duration": 2.656, "text": "the cool thing about these guys is that they have really..." }
  ]
}

With outputFormat: "srt" you get an srt string instead of segments. With "text" you get only text. publishDate is best effort and can be null.

A video without a transcript:

{
  "videoId": "oNTaHYXQ4MM",
  "url": "https://www.youtube.com/watch?v=oNTaHYXQ4MM",
  "title": "Humans Can Regrow an Organ",
  "channel": "Kurzgesagt – In a Nutshell",
  "errorCode": "NO_TRANSCRIPT",
  "error": "No captions/transcripts are available for this video."
}

Error codes: NO_TRANSCRIPT, VIDEO_UNAVAILABLE, AGE_RESTRICTED, LOGIN_REQUIRED (private or members-only), BLOCKED (YouTube kept blocking after all retries), FAILED (another error after all retries).

Pricing

This Actor uses pay-per-result pricing: you pay only for rows that contain a transcript. Videos without captions, and private, deleted, age-restricted or blocked videos, still appear in the dataset with an error field so you can see what happened, and they're free. Set a maximum spend per run and the Actor stops cleanly when it reaches it.

FAQ

Is it legal to scrape YouTube transcripts? The Actor only reads publicly available captions. Check YouTube's terms and the laws that apply to you, and respect the copyright of the creators.

Do I need a YouTube API key? No.

Why residential proxy, and will I get blocked? YouTube blocks scraping from datacenter IPs, the main reason transcript scrapers fail. This Actor defaults to Apify residential proxy. It uses the same method as the open-source youtube-transcript-api Python library (YouTube's internal player API with the Android client), which doesn't need a PO token for captions today. It detects blocks (HTTP 429, CAPTCHA pages, "confirm you're not a bot") and retries on a fresh IP, up to 6 times by default. It treats "this video has no transcript" as final, without retries or charges.

Can I use this to feed an LLM? Yes. Use outputFormat: "text" for the most compact output, or segments when you need timestamps for citations or chapter links.

What if YouTube changes something? YouTube changes its internals often, and runs may fail until the Actor is updated. The Actor always reports failed videos in the dataset and never drops them silently.

What are the limits?

• Public videos only. The Actor never logs in, so it can't read private, members-only or age-restricted videos.
• A transcript exists only if the uploader added captions or YouTube generated them. Some videos (music, very new or very short uploads) have none.
• Translations are YouTube's machine translations, not human ones.
• Channel expansion reads the channel's uploads list, which includes Shorts and past live streams.