YouTube Scraper – Videos, Comments & Transcripts (No API Key)

YouTube Scraper extracts structured data from YouTube without an API key or login: video metadata, comments (with replies), and transcripts/captions — by video URL, video ID, or keyword search. Export to JSON, CSV, or Excel.

This is a no-API-key YouTube comments API, YouTube transcript scraper, and video-metadata scraper in one Actor. Built for content researchers, marketers, data teams, and creators who need YouTube data at scale without the official Data API's quota limits.

What it does

• 🎬 Video metadata – title, description, channel name/ID/URL, view count, approximate like count, publish/upload date, duration, category, tags/keywords, thumbnails, and the list of available caption languages.
• 💬 Comments – author, text, like count, reply count, relative published time, pinned/hearted flags, and an avatar URL. Optionally expand reply threads.
• 📝 Transcripts – full transcript text plus timestamped segments, for manual or auto-generated captions, in your preferred language. Best-effort (see FAQ).
• 🔎 Keyword search – run a YouTube video search and automatically enrich every matching video.
• 🌍 Language control – pick the caption language and prefer human captions over auto-generated ones.
• 📤 Export anywhere – download the dataset as JSON, CSV, or Excel, or pull it via the Apify API.

Why this scraper

• No API key, no login, no quota. Uses YouTube's internal InnerTube endpoints (the same JSON APIs the website calls), reading the live client config from the watch page so the client version is never stale.
• Honest transcripts. Captions are gated by YouTube; this Actor uses the most reliable path available (InnerTube get_transcript via the iOS client, with a timedtext fallback) and is transparent about when a transcript can't be obtained.
• Built for reliability. Residential proxy by default, fresh proxy session per video, retries with jitter, and resumable state so a restarted run does not refetch finished videos.
• One Actor, four data types. Every record carries a type field (video, comment, transcript, searchResult) so you can split the dataset cleanly.
• Pairs with our other scrapers for social/video data (TikTok, Instagram, Reddit, and more) when you need cross-platform coverage.

How it works (and why it's reliable)

1. Seed – fetches the watch page once per video and extracts the live INNERTUBE config plus the initial comments token.

2. Metadata – read straight from the player response (no extra request).

3. Transcript – resolves the caption track list (language + manual/auto) from the mobile player response, then fetches the text in two stages:

   • Primary: the InnerTube get_transcript (transcript-panel) endpoint, called with the iOS mobile client. The params is a base64-protobuf built from the video ID + chosen track; this endpoint still returns segment text where the plain caption URL does not.
   • Fallback: the timedtext caption baseUrl (fmt=json3). The desktop/web client returns empty caption bodies (exp=xpe / PoToken gate), so mobile clients are used on purpose, but even mobile caption URLs can be gated on some IPs.

   Transcripts are therefore best-effort: when both paths are gated for a given video/IP, no transcript item is emitted (the video and comments are unaffected). A residential proxy materially improves the hit rate.

4. Comments – paginates the InnerTube next endpoint, joining the thread structure with the frameworkUpdates entity payloads to recover the real text/author/likes.

⚠️ Use a residential proxy. YouTube rate-limits datacenter IPs quickly (HTTP 429s and empty responses), especially for comment pagination. The default proxy group is RESIDENTIAL for this reason — and residential bandwidth is the main cost driver of a run.

Example input

{
  "videoUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],
  "searchQueries": ["lofi hip hop"],
  "maxResultsPerSearch": 25,
  "scrapeVideoMetadata": true,
  "scrapeTranscript": true,
  "scrapeComments": true,
  "maxComments": 100,
  "includeCommentReplies": false,
  "commentsSortBy": "top",
  "language": "en",
  "preferManualCaptions": true,
  "emitSearchResults": false,
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Example output

Each item carries a type field so you can split the dataset by record type. A video that is private, members-only, age-restricted, region-blocked, or not playable is emitted as a type: "skipped" item (with a reason) instead of crashing the run.

Video (type: "video")

{
  "type": "video",
  "videoId": "dQw4w9WgXcQ",
  "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "title": "Rick Astley - Never Gonna Give You Up (Official Music Video)",
  "description": "The official video for \"Never Gonna Give You Up\" by Rick Astley...",
  "channelName": "Rick Astley",
  "channelId": "UCuAXFkgsw1L7xaCfnd5JJOw",
  "channelUrl": "https://www.youtube.com/channel/UCuAXFkgsw1L7xaCfnd5JJOw",
  "viewCount": 1600000000,
  "likeCount": 17000000,
  "publishDate": "2009-10-25",
  "uploadDate": "2009-10-25",
  "durationSeconds": 212,
  "category": "Music",
  "keywords": ["rick astley", "never gonna give you up"],
  "isLiveContent": false,
  "isFamilySafe": true,
  "thumbnails": [{ "url": "https://i.ytimg.com/...", "width": 1280, "height": 720 }],
  "availableCaptionLanguages": [{ "languageCode": "en", "name": "English", "isAutoGenerated": false }],
  "commentCountApprox": 2200000,
  "foundViaQuery": null,
  "scrapedAt": "2026-06-26T10:00:00.000Z"
}

Transcript (type: "transcript")

{
  "type": "transcript",
  "videoId": "dQw4w9WgXcQ",
  "language": "en",
  "isAutoGenerated": false,
  "trackName": "English",
  "fullText": "We're no strangers to love ...",
  "segments": [{ "start": 18.5, "dur": 3.2, "text": "We're no strangers to love" }],
  "scrapedAt": "2026-06-26T10:00:00.000Z"
}

Comment (type: "comment")

{
  "type": "comment",
  "videoId": "dQw4w9WgXcQ",
  "commentId": "UgxKREWxIgDrw8w2e_Z4AaABAg",
  "text": "Still a banger in 2026.",
  "authorName": "@someuser",
  "authorChannelId": "UCabc...",
  "authorChannelUrl": "https://www.youtube.com/channel/UCabc...",
  "authorThumbnail": "https://yt3.ggpht.com/...",
  "likeCount": 1200,
  "replyCount": 8,
  "publishedTimeText": "3 weeks ago",
  "isPinned": false,
  "isHearted": false,
  "isReply": false,
  "parentCommentId": null,
  "scrapedAt": "2026-06-26T10:00:00.000Z"
}

Search result (type: "searchResult", only when Also emit raw search results is enabled)

{
  "type": "searchResult",
  "query": "lofi hip hop",
  "videoId": "jfKfPfyJRdk",
  "url": "https://www.youtube.com/watch?v=jfKfPfyJRdk",
  "title": "lofi hip hop radio",
  "channelName": "Lofi Girl",
  "channelId": "UCSJ4gkVC6NrvII8umztf0Ow",
  "viewCountText": "Streaming",
  "publishedTimeText": null,
  "lengthText": null,
  "thumbnails": [{ "url": "https://i.ytimg.com/...", "width": 360, "height": 202 }],
  "scrapedAt": "2026-06-26T10:00:00.000Z"
}

Output fields

Video (type: "video")

Field	Type	Description
type	string	Always "video".
videoId	string	11-character video ID.
url	string	Canonical watch URL.
title	string	Video title.
description	string	Full description text.
channelName	string	Channel (author) name.
channelId	string	Channel ID.
channelUrl	string	Channel URL.
viewCount	integer	Total views (nullable).
likeCount	integer	Approximate like count from the watch UI (nullable; YouTube does not expose an exact number).
publishDate	string	Publish date from microformat (nullable).
uploadDate	string	Upload date from microformat (nullable).
durationSeconds	integer	Duration in seconds (nullable).
category	string	Video category (nullable).
keywords	array	Tags array (may be empty).
isLiveContent	boolean	Whether the video is/was live (nullable).
isFamilySafe	boolean	Family-safe flag (nullable).
thumbnails	array	Array of { url, width, height }.
availableCaptionLanguages	array	Array of { languageCode, name, isAutoGenerated }.
commentCountApprox	integer	Approximate comment count from the header (nullable).
foundViaQuery	string	The search query that surfaced this video (only present for videos discovered via searchQueries).
scrapedAt	string	ISO 8601 timestamp.

Transcript (type: "transcript")

Field	Type	Description
type	string	Always "transcript".
videoId	string	Video ID.
language	string	Caption language code (nullable).
isAutoGenerated	boolean	true for auto (ASR) captions.
trackName	string	Human-readable track name (nullable).
fullText	string	All segments joined into one string.
segments	array	Array of { start, dur, text } (seconds as floats; start/dur may be null).
scrapedAt	string	ISO 8601 timestamp.

Comment (type: "comment")

Field	Type	Description
type	string	Always "comment".
videoId	string	Video the comment belongs to.
commentId	string	Comment ID.
text	string	Comment body.
authorName	string	Author handle / display name.
authorChannelId	string	Author channel ID (nullable).
authorChannelUrl	string	Author channel URL (nullable).
authorThumbnail	string	Author avatar URL (nullable).
likeCount	integer	Like count (defaults to 0 when not shown).
replyCount	integer	Reply count (defaults to 0).
publishedTimeText	string	Relative time string (e.g. "3 weeks ago"); YouTube does not expose absolute comment dates.
isPinned	boolean	Whether the comment is pinned.
isHearted	boolean	Whether the creator hearted it.
isReply	boolean	true for replies.
parentCommentId	string	Parent comment ID (null for top-level).
scrapedAt	string	ISO 8601 timestamp.

Search result (type: "searchResult")

Field	Type	Description
type	string	Always "searchResult".
query	string	The search query that matched.
videoId	string	Video ID.
url	string	Watch URL.
title	string	Video title.
channelName	string	Channel name (nullable).
channelId	string	Channel ID (nullable).
viewCountText	string	View-count display string from the search card (nullable).
publishedTimeText	string	Relative publish-time display string (nullable).
lengthText	string	Duration display string (nullable).
thumbnails	array	Array of { url, width, height }.
scrapedAt	string	ISO 8601 timestamp.

Skipped (type: "skipped")

Field	Type	Description
type	string	Always "skipped".
videoId	string	Video ID that could not be scraped.
url	string	Watch URL.
reason	string	Why it was skipped (e.g. LOGIN_REQUIRED, AGE_VERIFICATION_REQUIRED, no_video_details, watch_fetch_failed: ...).
scrapedAt	string	ISO 8601 timestamp.

Input reference

Field	Type	Default	Description
videoUrls	array	–	YouTube watch URLs (watch?v=, youtu.be/, /shorts/, /embed/, /live/). Normalized to the 11-char ID.
videoIds	array	–	Raw 11-character video IDs.
searchQueries	array	–	Keyword search strings (videos only); matches are enriched like direct inputs.
maxResultsPerSearch	integer	50	Max videos per search query (pagination stops once reached).
scrapeVideoMetadata	boolean	true	Emit type: "video" items.
scrapeTranscript	boolean	true	Fetch transcript/captions (best-effort).
scrapeComments	boolean	false	Fetch comments (slowest, most rate-limited path).
maxComments	integer	100	Max top-level comments per video; 0 = unlimited.
includeCommentReplies	boolean	false	Also expand reply threads.
commentsSortBy	string	top	"top" or "newest".
language	string	en	Preferred caption language code; also sets the InnerTube hl locale.
preferManualCaptions	boolean	true	Prefer human captions over auto-generated.
emitSearchResults	boolean	false	Also push raw type: "searchResult" items for searches.
proxyConfiguration	object	RESIDENTIAL	Apify proxy; RESIDENTIAL strongly recommended (and the default).

Common use cases

• Transcript mining – pull YouTube transcripts for summarization, search, or LLM/RAG pipelines.
• Comment & sentiment analysis – export top or newest comments (and replies) as a YouTube comments API for NLP and analysis.
• Channel & topic research – search a keyword and collect metadata across many videos at once.
• Content monitoring – track view counts, approximate likes, and comment volume over time.
• Competitor & market research – build datasets from competitor channels and export to JSON/CSV/Excel.

FAQ

Do I need a YouTube/Google API key? No. This YouTube scraper uses YouTube's internal InnerTube endpoints — no API key, no OAuth, no quota.

Why are some transcripts missing? Transcripts are best-effort. Not every video has captions, and some only have auto-generated ones. YouTube also gates caption text behind PoToken/IP checks: the Actor first tries the InnerTube get_transcript panel endpoint via the iOS client (which still returns text on mobile clients) and falls back to the timedtext caption URL, but on a gated video/IP combination both can come back empty — in that case the video and comments are still scraped, just without a transcript. A residential proxy improves the YouTube transcript hit rate.

Why are comments slow, and why do they need a proxy? Comments are the slowest, most rate-limited path: the Actor paginates the InnerTube next endpoint one page at a time with jitter to stay under YouTube's limits, so a video with many comments takes proportionally longer. Datacenter IPs get 429s and empty responses quickly, so a residential proxy is strongly recommended when scraping comments at volume. Comments are off by default for this reason.

Why do I need a residential proxy, and what does it cost? Datacenter IPs get rate-limited (429s) and served empty responses, especially during comment pagination, so the default proxy group is RESIDENTIAL. Residential proxy bandwidth is the main cost driver of a run — fetching watch pages, comments, and transcripts all consume residential traffic — so scope your maxComments and maxResultsPerSearch to control spend.

Which languages are supported? Any language YouTube has a caption track for. Set language to a language code (e.g. en, es, de, fr, pt, hi, ja) to pick the caption track and the InnerTube hl locale; if that exact track is missing, the scraper falls back to another track or an auto-generated one. preferManualCaptions controls whether human captions win over ASR.

How are limits and pagination handled? maxResultsPerSearch caps videos per search query, and maxComments caps top-level comments per video (0 = unlimited). Search and comments paginate automatically until the limit is hit or YouTube stops returning a continuation. Runs are resumable: a restarted run skips videos it already finished.

What export formats are available? The dataset can be exported to JSON, CSV, or Excel (also XML, JSONL, RSS, or HTML table) from the Apify Console, or fetched programmatically via the Apify API.

---

This YouTube scraper pairs well with our other social and video scrapers (TikTok, Instagram, Reddit, and more) when you need cross-platform data.