YouTube Transcript & Subtitles Scraper API for AI | $4.50/1k

What does YouTube Transcript Scraper do?

YouTube Transcript Scraper gets the transcript of any public YouTube video as clean text: the creator's own subtitles when they exist, otherwise YouTube's auto-generated captions. Paste video links or IDs, or a channel or playlist link, and get one row per video with the full transcript, its language, whether it is auto-generated, a word count and, if you want them, timestamped caption lines. It works as a YouTube transcript API: no YouTube API key, no Google account, no quota and no browser.

Use it to:

• feed LLM pipelines and RAG: summarise videos, answer questions over a channel's back catalogue, or build a searchable knowledge base;
• turn videos into content: blog posts, show notes, newsletters, social posts and SEO articles;
• do research and monitoring: what a competitor, analyst or channel said, searched as text;
• create datasets for NLP, translation and speech research from real spoken language;
• add subtitles or quotes to your own workflow with exact timestamps.

Why use YouTube Transcript Scraper?

• No API key, no quota. YouTube's Data API does not hand out caption text for videos you do not own; this Actor needs no key at all.
• Pick the language. Ask for en, es, de, pt-BR or any other code. A creator-made track wins, then the auto-generated one; every row also lists all tracks the video has in availableLanguages.
• Search by keyword. Type a keyword and get the transcripts of the top 10, 50 or 500 videos YouTube finds for it, sorted by relevance, upload date or views, filtered by upload date and length. Each row says which keyword found it and at what rank.
• Channels and playlists too. Paste @handle, a channel URL or a playlist URL and get the latest videos' transcripts in one run.
• Plain text, timestamps, SRT or VTT. transcript is always one clean string; switch on includeTimestamps for every caption line with its start time and duration, or pick Subtitle files to get ready-to-upload .srt and .vtt text on every row at no extra cost.
• Video metadata included. Views, description, tags, thumbnail, length and channel come with every transcript, so you can filter and rank without a second scraper.
• Honest, free "no captions" rows. A video without captions, a private or a removed video returns one free not_found row that says why. You pay only for transcripts.
• Built for scale. Every request goes through a fresh residential IP, so large batches keep working. The default run of 5 videos finishes in under a minute.

What data can YouTube Transcript Scraper extract?

Field	Type	Description
videoId	string	The video's 11-character id.
url	string	The video's watch URL.
title	string	Video title.
channelName	string	Name of the channel that published the video.
channelId	string	Channel id (UC...).
language	string	Language code of the returned transcript, e.g. en.
languageName	string	Track name as YouTube shows it, e.g. English (auto-generated).
isAutoGenerated	boolean	true for YouTube's automatic speech-recognition captions, false for creator-made subtitles.
availableLanguages	array	Every caption track of the video: languageCode, name, isAutoGenerated.
transcript	string	The full transcript as one plain-text string.
wordCount	integer	Words in the transcript.
segmentCount	integer	Caption lines in the transcript.
segments	array	With Include timestamps: { start, duration, text } per caption line, in seconds.
durationSeconds	integer	Video length in seconds.
viewCount	integer	Views at the time of the run.
description	string	Video description (email addresses, phone numbers, social profile links and @handles are removed).
keywords	array	The creator's tags.
thumbnailUrl	string	Largest thumbnail image.
isLiveContent	boolean	true for live streams and their replays.
srt / vtt	string	With Subtitle files: the transcript as an SRT or WebVTT file.
searchQuery, searchRank, publishedTimeText	string, integer, string	For videos found by Search keywords: the keyword, the position in YouTube's results (1 = top) and the upload age as YouTube shows it (e.g. 2 years ago).

Result status (tri-state output)

status	Meaning	Billed?
ok	One video's transcript.	Yes
not_found	The video has no captions, or YouTube does not show it to logged-out visitors (private, removed, age-restricted), or a keyword found no videos. error says which; the video's metadata is still included.	No
error	The video could not be read after several retries, or an input line was not a YouTube link. error says why.	No

How to scrape YouTube transcripts with YouTube Transcript Scraper

1. Open the Actor in Apify Console and go to the Input tab.
2. Paste YouTube video URLs or IDs (watch links, youtu.be links, Shorts, live links or bare 11-character IDs), channel and playlist links, or type Search keywords.
3. Set the Preferred language, switch on Include timestamps or pick Subtitle files if you need them, and choose Videos per keyword or Videos per channel or playlist.
4. Click Start. The default input (5 videos) finishes in under a minute.
5. Download the transcripts as JSON, CSV, Excel or HTML, or read them through the API.

Need richer filters (likes, exact upload dates, comments)? List the videos with YouTube Scraper and pass their URLs here.

How much does it cost to scrape YouTube transcripts?

The Actor uses pay-per-event pricing: you pay for the transcripts you get, Apify platform usage and residential proxy traffic are included.

Event	Price
Actor start	$0.005 per run
Result (status: ok row: one video's transcript)	$0.0045 per transcript ($4.50 per 1,000)

Example: 1,000 transcripts cost about $4.51 ($4.50 plus the $0.005 start fee). Long videos cost the same as short ones, and searching, metadata and SRT/VTT files cost nothing extra. Rows with status not_found or error are never billed. You can cap spending per run with Maximum results and in the run's Max total charge option.

Input

Field	Type	Default	Description
startUrls	array of strings	five sample videos	Video URLs or IDs; channel URLs, @handles or UC... ids; playlist URLs or ids; YouTube search URLs. One per line. Left at the example, it is skipped when you give searchQueries.
searchQueries	array of strings	none	Keywords; the top videos YouTube finds for each are transcribed.
maxVideosPerSearch	integer	10	Top search results taken per keyword (1 to 500).
searchSort	string	relevance	relevance, date (upload date), views or rating.
uploadDate	string	any	any, hour, today, week, month or year.
videoDuration	string	any	any, short (under 4 min), medium (4-20 min) or long (over 20 min).
subtitleFormats	array of strings	none	srt and/or vtt: add the transcript as subtitle files.
language	string	en	Preferred transcript language code. Falls back to YouTube's default track when the video has no track in this language.
includeTimestamps	boolean	false	Add the segments array (start, duration, text per caption line).
maxVideosPerSource	integer	10	Latest videos taken from each channel or playlist link (1 to 100).
maxItems	integer	20	Total rows; each ok row is one billable transcript.

Example input (the prefill):

{
    "startUrls": [
        "https://www.youtube.com/watch?v=jNQXAC9IVRw",
        "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
        "https://www.youtube.com/watch?v=RcGyVTAoXEU",
        "https://www.youtube.com/watch?v=0NY2gAftzJE",
        "https://www.youtube.com/watch?v=kJQP7kiw5Fk"
    ],
    "language": "en",
    "maxItems": 20
}

The latest 25 videos of a channel in Spanish where available, with timestamps:

{
    "startUrls": ["@TED"],
    "language": "es",
    "includeTimestamps": true,
    "maxVideosPerSource": 25,
    "maxItems": 25
}

The transcripts of the 50 most-viewed long videos about a topic from this year, with SRT files:

{
    "searchQueries": ["kubernetes tutorial"],
    "maxVideosPerSearch": 50,
    "searchSort": "views",
    "uploadDate": "year",
    "videoDuration": "long",
    "subtitleFormats": ["srt"],
    "maxItems": 50
}

Output

A real row from the default run (transcript and track list shortened):

{
    "url": "https://www.youtube.com/watch?v=RcGyVTAoXEU",
    "status": "ok",
    "scrapedAt": "2026-10-19T07:07:27.425Z",
    "videoId": "RcGyVTAoXEU",
    "title": "How to Make Stress Your Friend | Kelly McGonigal | TED",
    "channelName": "TED",
    "channelId": "UCAuUUnT6oDeKwE6v1NGQxug",
    "durationSeconds": 869,
    "language": "en",
    "languageName": "English",
    "isAutoGenerated": false,
    "availableLanguages": [
        { "languageCode": "sq", "name": "Albanian", "isAutoGenerated": false },
        { "languageCode": "en", "name": "English", "isAutoGenerated": false },
        { "languageCode": "en", "name": "English (auto-generated)", "isAutoGenerated": true }
    ],
    "transcript": "I have a confession to make. But first, I want you to make a little confession to me. In the past year, I want you to just raise your hand if you've experienced relatively little stress ...",
    "wordCount": 2012,
    "segmentCount": 284
}

With Include timestamps on, each row also has a segments array (first line of video jNQXAC9IVRw shown):

"segments": [
    { "start": 1.2, "duration": 2.16, "text": "All right, so here we are, in front of the elephants" }
]

Every row also carries the video's viewCount, description, keywords, thumbnailUrl and isLiveContent; rows found by a keyword search add searchQuery, searchRank and publishedTimeText. With Subtitle files set to SRT, srt holds the file:

"srt": "1
00:00:01,200 --> 00:00:03,360
All right, so here we are, in front of the elephants

2
..."

A video without captions (free):

{
    "url": "https://www.youtube.com/watch?v=oNTaHYXQ4MM",
    "status": "not_found",
    "error": "This video has no captions or subtitles (free row, not charged).",
    "videoId": "oNTaHYXQ4MM",
    "title": "Humans Can Regrow an Organ",
    "channelName": "Kurzgesagt – In a Nutshell",
    "durationSeconds": 74
}

Tips

• For LLMs: transcript is a single clean string, ready to chunk and embed; wordCount helps you estimate tokens before you send it anywhere.
• Creator subtitles first: filter on isAutoGenerated: false when punctuation and spelling matter; auto-generated captions are lowercase-heavy and unpunctuated.
• Multilingual channels: check availableLanguages to see every track, then rerun with language set to the one you need.
• Whole channels: a channel link takes its newest uploads (videos, Shorts and live replays), up to 100 per link.
• Research a topic: search a keyword with searchSort: views and videoDuration: long to get the most-watched in-depth videos; searchRank keeps YouTube's order.
• Subtitles for editing: srt and vtt open directly in video editors and players; save the field's text as a .srt or .vtt file.

Limitations

• Only captions that exist on YouTube are returned. Videos without any caption track (common for music, very new uploads and some Shorts) give a free not_found row; the Actor does not transcribe audio itself.
• No translation: you get the tracks the video has, not YouTube's machine-translated versions (YouTube refuses automated translation requests).
• Keyword search takes up to 500 videos per keyword, in YouTube's order for a logged-out visitor in the US. A video found by two keywords is transcribed (and billed) once.
• Channel and playlist links take up to 100 of the latest videos per link.
• Private, removed, members-only and age-restricted videos are not available to logged-out visitors and return a free not_found row.
• Results reflect what YouTube shows a logged-out visitor at the time of the run; if YouTube changes its data, rows come back as error and the Actor is updated quickly (report it in the Issues tab).

FAQ

Do I need a YouTube API key or a Google account?

No. The Actor reads the same caption tracks the YouTube player loads for a logged-out visitor. There is no API key, no quota and nothing to sign in to.

Which transcript do I get when a video has several?

A creator-made track in your Preferred language, else the auto-generated track in that language, else YouTube's default track for the video (usually the spoken language). language, languageName and isAutoGenerated tell you what you got, and availableLanguages lists every track.

Does a 3-hour podcast cost more than a 1-minute Short?

No. Every transcript is one result at $0.0045, whatever its length.

Is it legal to scrape YouTube transcripts?

Scraping publicly available data is generally legal, but you are responsible for how you use the output. Transcripts are the creators' content and may be protected by copyright: use them for analysis, research, search and other purposes the law allows, and get permission before republishing them. Read the legal notice below and YouTube's terms of service.

Can I use this Actor from an AI agent or MCP client?

Yes. Call it through the Apify API or the Apify MCP server with a list of video URLs; every row has a fixed set of fields and a status, so an agent can tell "no captions" from "could not read".

Why did I get fewer rows than maxItems?

maxItems is a ceiling. You get fewer rows when you pass fewer videos, when a channel or playlist has fewer videos than Videos per channel or playlist, or when that setting times the number of links is below maxItems.

What if a field is missing?

Optional fields are omitted when YouTube does not provide them (for example title on a removed video). If a field that should always be present is missing, open an issue in the Issues tab with the run ID.

Related Actors

• YouTube Scraper - search YouTube, list a channel's videos or a playlist with views, likes and durations, then get their transcripts here.
• YouTube Comments Scraper - add what viewers said under the same videos.
• Google Trends Scraper - check whether the topics in the transcripts are rising or fading.
• Keyword Suggestions API - turn the themes you find into the searches people actually type.

Legal and data-protection notice

This Actor extracts only data that YouTube publishes publicly to logged-out visitors; it does not extract private user data such as e-mail addresses, phone numbers, gender or precise location, and it does not log in or get around access controls. Transcripts are the creators' content and may be protected by copyright; spoken words can also mention people, so a transcript could contain personal data. Personal data is protected by the GDPR in the European Union and by other regulations around the world. You should not scrape personal data unless you have a legitimate reason to do so. If you are unsure whether your reason is legitimate, consult your lawyers. You are responsible for complying with YouTube's terms of service, copyright and applicable law when using the extracted data.

This Actor is an independent tool and is not affiliated with, endorsed by or sponsored by YouTube or Google. All trademarks belong to their respective owners.