YouTube Transcript Scraper: Subtitles, Channels, Search avatar

YouTube Transcript Scraper: Subtitles, Channels, Search

Pricing

from $1.10 / 1,000 transcripts

Go to Apify Store
YouTube Transcript Scraper: Subtitles, Channels, Search

YouTube Transcript Scraper: Subtitles, Channels, Search

Extract YouTube transcripts and subtitles in bulk from videos, Shorts, playlists, channels and search. Timestamped segments, plain text, SRT and VTT, auto-generated flag, all caption languages, plus title, channel, publish date, views and duration. Only successful transcripts are charged.

Pricing

from $1.10 / 1,000 transcripts

Rating

0.0

(0)

Developer

ZH Tech

ZH Tech

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

2 days ago

Last modified

Features

Limited permissions

Share

Extract YouTube transcripts, subtitles and captions in bulk: single videos, Shorts, whole playlists, entire channels and YouTube search results, in one run. Every row comes with timestamped segments, clean plain text, optional SRT and VTT subtitle files, the caption language, whether it is auto-generated, every caption language the video offers, and full video metadata including the exact publish date, duration, views and likes.

You only pay for transcripts that were actually delivered. Videos without captions, private or deleted videos and anything else that fails come back as free rows with a clear error code, so you always know why.

Built and maintained by ZH Tech.

What can this YouTube Transcript Scraper do?

  • Every kind of input in one field: watch links, youtu.be links, Shorts, live and embed links, bare video IDs, playlists, channels (@handle, /channel/UC..., /c/, /user/, or a channel tab such as /@name/shorts) and YouTube search URLs. Add plain search queries in a separate field. Duplicates are removed.
  • Channels and playlists: set how many videos to take per source, choose regular videos, Shorts and past live streams, and filter by publish date (absolute dates or relative periods like "30 days"). Channel listings stop early once older videos are reached, so you never pay for videos outside your range.
  • YouTube search: get transcripts for the top results of any query, sorted by relevance or popularity (view count).
  • Auto-generated and human captions: prefer human-written captions, require them, or always use YouTube's speech recognition track.
  • Smart language handling: list languages in order of preference (en, es, pt-BR). Base codes match regional tracks (pt matches pt-BR). If none exists, the actor falls back to the language the video was actually spoken in, detected from YouTube's audio track data, so you do not get an auto-dubbed Arabic transcript for an English video.
  • Every caption track listed: each row includes availableLanguages, so you can see what else a video offers.
  • Many formats: timestamped segments, plain text, timestamped paragraphs for LLM citations, SRT, WebVTT, and optional word-level timestamps.
  • Clean text option: strip [Music], [Applause] and music notes.
  • Scheduled monitoring: "Only new videos" remembers what was already transcribed, so a daily schedule on a channel returns only fresh uploads.
  • Built to avoid blocking: Apify Proxy with session rotation, retries, and automatic escalation from datacenter to residential IPs only for the requests YouTube blocks. Several extraction methods are tried before a video is given up on, and "no captions" is double checked from a second IP before it is reported.
  • Fast: a few videos per second with default settings, hundreds of videos in a couple of minutes.
  • Never crashes on a bad video: each problem becomes a free row with an errorCode.

Use cases

  • AI, RAG and LLM pipelines: feed transcripts into a vector database, chatbot knowledge base or fine-tuning dataset. The timestampedText format keeps [m:ss] markers so an LLM can cite the exact moment in a video.
  • Content repurposing: turn videos into blog posts, newsletters, social threads, show notes and quote cards.
  • Research and analysis: study what creators, experts or competitors say across hundreds of videos. Filter a channel by date and compare topics over time.
  • SEO and marketing: find the keywords and questions the top videos in your niche talk about.
  • Accessibility and localization: download SRT or VTT subtitles to edit, translate or re-upload.
  • Monitoring: schedule a daily run on a list of channels with "Only new videos" and push new transcripts to Slack, Google Sheets, Notion or your own API.

How to use it

  1. Click Try for free.
  2. Paste YouTube URLs (videos, Shorts, playlists or channels) into YouTube URLs or IDs, and/or add Search queries.
  3. Optionally set Max videos per channel, playlist or search, a Published after date and your Preferred transcript languages.
  4. Click Start. Results appear in the Output tab within seconds. Download them as JSON, CSV, Excel, XML or HTML, or fetch them via the API.

Input examples

Transcripts of a few videos:

{
"urls": [
"https://www.youtube.com/watch?v=aircAruvnKk",
"https://youtu.be/dQw4w9WgXcQ",
"https://www.youtube.com/shorts/V_uq3v6xK-4"
]
}

The last 30 days of a channel, regular videos only, as plain text and SRT:

{
"urls": ["https://www.youtube.com/@3blue1brown"],
"channelContentTypes": ["videos"],
"publishedAfter": "30 days",
"maxVideosPerSource": 50,
"outputFormats": ["text", "srt"]
}

Top 20 search results in Spanish, falling back to the original language when Spanish is missing:

{
"searchQueries": ["receta de pan casero"],
"maxVideosPerSource": 20,
"languages": ["es"],
"fallbackToAnyLanguage": true
}

Daily channel monitoring (schedule this input; each run returns only new uploads):

{
"urls": ["https://www.youtube.com/@veritasium", "https://www.youtube.com/@mkbhd"],
"maxVideosPerSource": 10,
"onlyNewVideos": true,
"stateKey": "tech-channels"
}

All input options

FieldDescriptionDefault
urlsVideos, Shorts, playlists, channels, search URLs or bare IDs
searchQueriesYouTube search queries
maxVideosPerSourceVideos per channel, playlist or search (0 = no limit)20
maxTranscriptsStop after this many successful transcripts (0 = no limit)0
channelContentTypesvideos, shorts, streamsall three
publishedAfter / publishedBeforeDate (2026-01-31) or relative (30 days). Applies to channels, playlists and searches
searchSortrelevance or views (popularity). YouTube no longer sorts searches by upload date or rating: old inputs with date run as relevance, rating as views, with a log note. For recent videos use publishedAfterrelevance
languagesPreferred caption languages in order["en"]
fallbackToAnyLanguageReturn the original spoken language when none of yours existtrue
captionTypeany (prefer human), manual, autoany
translateToAsk YouTube for a machine translation (best effort, see limitations)
outputFormatssegments, text, timestampedText, srt, vttsegments, text
includeWordTimestampsPer-word start times inside segmentsfalse
removeSoundTagsRemove [Music], [Applause] and similarfalse
includeDescriptionAdd description and tagsfalse
includeErrorItemsAdd free rows for videos without a transcripttrue
onlyNewVideos / stateKeySkip videos transcribed in earlier runsfalse / default
proxyConfigurationApify Proxy or your own proxiesApify Proxy
residentialFallbackRetry blocked requests through residential IPstrue
maxConcurrencyVideos processed in parallel10
maxRetriesRetries per blocked request5

Output example

{
"videoId": "aircAruvnKk",
"url": "https://www.youtube.com/watch?v=aircAruvnKk",
"status": "success",
"errorCode": null,
"title": "But what is a neural network? | Deep learning chapter 1",
"channelName": "3Blue1Brown",
"channelId": "UCYO_jab_esuFRV4b17AJtAw",
"channelHandle": "@3blue1brown",
"channelUrl": "https://www.youtube.com/channel/UCYO_jab_esuFRV4b17AJtAw",
"publishedAt": "2017-10-05T08:11:25-07:00",
"durationSeconds": 1120,
"duration": "18:40",
"viewCount": 25275345,
"likeCount": 570600,
"category": "Education",
"isShort": false,
"isLiveContent": false,
"isFamilySafe": true,
"thumbnailUrl": "https://i.ytimg.com/vi/aircAruvnKk/hqdefault.jpg",
"videoLanguage": null,
"language": "en",
"languageName": "English",
"isAutoGenerated": false,
"isLanguageFallback": false,
"isFromAutoDub": false,
"isTranslated": false,
"translatedFrom": null,
"translationStatus": null,
"availableLanguages": [
{ "code": "en", "name": "English", "isAutoGenerated": false, "isTranslatable": true },
{ "code": "en", "name": "English (auto-generated)", "isAutoGenerated": true, "isTranslatable": true },
{ "code": "es", "name": "Spanish", "isAutoGenerated": false, "isTranslatable": true }
],
"segmentCount": 286,
"wordCount": 3357,
"text": "This is a 3. It's sloppily written and rendered at an extremely low resolution of 28x28 pixels, but your brain has no trouble recognizing it as a 3. ...",
"segments": [
{ "start": 4.22, "end": 5.4, "duration": 1.18, "text": "This is a 3." },
{ "start": 6.06, "end": 10.713, "duration": 4.653, "text": "It's sloppily written and rendered at an extremely low resolution of 28x28 pixels," }
],
"extractionMethod": "innertube-android",
"input": "https://www.youtube.com/watch?v=aircAruvnKk",
"source": "video:aircAruvnKk",
"scrapedAt": "2026-11-16T02:37:12+00:00"
}

videoLanguage is the original spoken language when YouTube reports it (videos with several audio tracks), otherwise null. isFromAutoDub is true when the transcript comes from one of YouTube's AI-dubbed audio tracks rather than the original audio.

A video without a transcript (free row):

{
"videoId": "8plwv25NYRo",
"url": "https://www.youtube.com/watch?v=8plwv25NYRo",
"status": "error",
"errorCode": "NO_CAPTIONS",
"errorMessage": "This video has no captions or auto-generated transcript on YouTube. Not charged.",
"title": "10 Hours of Rain Sound Relaxation / Ultimate Stress Relief",
"availableLanguages": []
}

Error codes

CodeMeaning
NO_CAPTIONSThe video has no captions and no auto-generated transcript (common for music, ambient videos, very new uploads and many Shorts)
LANGUAGE_NOT_AVAILABLECaptions exist but not in your languages, and fallback is off. See availableLanguages
VIDEO_UNAVAILABLEDeleted, removed or never existed
PRIVATE_VIDEOPrivate video
AGE_RESTRICTEDRequires a signed-in adult account. This actor never logs in
MEMBERS_ONLYChannel membership content
LIVE_NOT_FINISHEDLive stream in progress, upcoming premiere, or a stream without a recording
GEO_BLOCKEDNot available in the proxy's country. Choose a proxy country where it is available
BLOCKEDYouTube kept blocking even after rotating IPs. Rare. Run again or use the RESIDENTIAL proxy group
FETCH_FAILEDUnexpected download error
LISTING_FAILED, CHANNEL_NOT_FOUNDA channel, playlist or search could not be listed
INVALID_INPUTThe input is not a YouTube URL or ID

Each run also saves a RUN_SUMMARY record in its key-value store with counts per error code, time taken and proxy statistics.

How much does it cost to scrape YouTube transcripts?

This actor uses pay per event pricing:

  • Transcript: charged once per video for which a transcript is returned (status: "success"). Rows with an error code are free.
  • Actor start: a tiny fixed fee per run.

The exact prices are on the Pricing tab. Platform usage (compute and proxy) is included, so the price you see is the price you pay. You can cap your spend with the "Maximum cost per run" setting: the actor checks the remaining budget before each transcript and stops cleanly instead of overspending.

Tips to keep costs low:

  • Use maxVideosPerSource and publishedAfter for channels. Old videos are skipped during listing and never charged.
  • Use onlyNewVideos for scheduled runs so you never pay twice for the same video.
  • Put many videos into one run instead of starting one run per video, to save the start fee.

Use with the API, integrations, LLMs and MCP

API

Run the actor and get the results in one request (replace YOUR_TOKEN):

curl -X POST "https://api.apify.com/v2/acts/zhtech~youtube-transcript-scraper/run-sync-get-dataset-items?token=YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{"urls": ["https://www.youtube.com/watch?v=aircAruvnKk"], "outputFormats": ["text"]}'

Python:

from apify_client import ApifyClient
client = ApifyClient("YOUR_TOKEN")
run = client.actor("zhtech/youtube-transcript-scraper").call(run_input={
"urls": ["https://www.youtube.com/@3blue1brown"],
"maxVideosPerSource": 10,
"outputFormats": ["text"],
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
if item["status"] == "success":
print(item["title"], len(item["text"]))

JavaScript:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: 'YOUR_TOKEN' });
const run = await client.actor('zhtech/youtube-transcript-scraper').call({
urls: ['https://youtu.be/dQw4w9WgXcQ'],
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items[0].text);

Integrations

Connect the actor to Make, Zapier, n8n, Google Sheets, Slack, Airtable, Google Drive, webhooks and more from the Integrations tab. A typical flow: schedule a daily run with onlyNewVideos, then send each new transcript to an LLM step that writes a summary into Notion or Slack.

Using it with LLMs and AI agents (MCP)

The actor works as a tool for AI agents through the Apify MCP server (mcp.apify.com). Add zhtech/youtube-transcript-scraper as a tool in Claude, Cursor, VS Code or any other MCP client, then ask things like "Summarize the last five videos of @veritasium" or "What does this video say about pricing?". For LLM prompts, request outputFormats: ["timestampedText"]: it is compact and keeps [m:ss] markers so the model can cite moments in the video.

For RAG, use the segments array to chunk by time (for example 60 second windows) and store videoId, start and url as metadata, so every answer can link to https://www.youtube.com/watch?v=VIDEO_ID&t=START.

FAQ

Does it work for videos with only auto-generated captions? Yes. Most YouTube videos only have auto-generated captions, and they are returned with isAutoGenerated: true. Set captionType to manual if you only want human-written captions.

Can I get transcripts for an entire channel? Yes. Paste the channel URL and set maxVideosPerSource to 0 for no limit, or to the number of newest videos you want. Choose videos, Shorts and past live streams with channelContentTypes.

Does it support YouTube Shorts? Yes, both single Shorts links and the Shorts tab of a channel. Rows include isShort. Note that many Shorts have no captions at all; those return a free NO_CAPTIONS row.

Which languages are supported? Every language YouTube has captions for. Each row lists all available tracks in availableLanguages.

What happens when a video has no transcript? You get a free row with errorCode: "NO_CAPTIONS". Nothing is charged. This actor reads YouTube's own captions and does not run speech-to-text on the audio.

Can it translate transcripts? If the video already has a track in your translateTo language, that track is used. Otherwise the actor asks YouTube for its automatic translation. At the time of writing YouTube refuses most automated translation requests, so treat this as best effort: you always get the original transcript, and translationStatus tells you what happened (translated, refused_by_youtube or not_offered).

Why are some requests retried? YouTube limits requests from datacenter IPs. The actor rotates IPs and switches individual requests to residential IPs when needed. You do not need to configure anything.

Can I scrape private, members-only or age-restricted videos? No. The actor never logs in to YouTube and only reads publicly available captions. Those videos return a free error row.

Is it legal to scrape YouTube transcripts? The actor only collects publicly available data. Transcripts can be protected by copyright, so check how you plan to use them and respect YouTube's terms and the rights of creators. If you are unsure, consult a lawyer.

Limitations

  • Only captions that exist on YouTube are returned. There is no speech-to-text for videos without captions.
  • YouTube machine translation (translateTo) is usually refused for automated clients. Native tracks in your language always work.
  • Live streams can be transcribed only after they end and YouTube has processed the captions. Very recent uploads may not have auto captions yet; with onlyNewVideos they are retried on the next run.
  • Age-restricted, private and members-only videos are not supported (no login).
  • publishedAfter and publishedBefore apply to channels, playlists and searches. Videos you list explicitly are always processed.
  • View and like counts are a snapshot at the time of the run.

Support

Found a problem or need a feature? Open an issue on the Issues tab with the run ID and the video URL.