Video Transcript API — YouTube to Text for Developers and AI

A developer-first video transcript API: send YouTube video URLs or IDs, get back JSON with the full text and timestamped segments. Built for LLM summarisation, RAG pipelines, AI agents and backend jobs — as a batch run, a live HTTP endpoint or an MCP tool.

What does this video transcript API do?

Give it YouTube video links or IDs. For every video it returns:

• the full transcript as one clean string, ready for an LLM, a search index or a spreadsheet;
• timestamped segments (start, duration, text) for subtitles, players or chunked RAG;
• the language, whether captions are human-written or auto-generated, and every other caption language available;
• the video title, channel, duration, word count and segment count.

Watch links, youtu.be short links, Shorts, embeds and bare 11-character IDs all work, mixed in one list.

Why use it

• One predictable JSON shape for every video, with a per-video error instead of a crashed run.
• One HTTP call — the run-sync endpoint returns the transcripts as JSON, ideal for a backend or an agent.
• MCP tool video_transcript — plug it into Claude, Cursor or any MCP client.
• No keys, no cookies, no OAuth — nothing to rotate.
• $0.005 per transcript, failures free, with a hard spending cap per run.

How to get YouTube transcripts in 3 steps

1. Click Try for free and paste one or more YouTube links in YouTube videos.
2. Optional: set a Preferred language (en, fr, es, de, …). Leave it empty for English, then the video's default.
3. Click Start. Download the results as JSON, CSV, Excel, HTML or XML, or read them from the API.

Input

Field	Type	What it does
videos	array	YouTube watch URLs, youtu.be links, Shorts, embeds or 11-character video IDs. Up to 1,000 per run.
language	string	Preferred two-letter caption language. Human-written captions are always preferred over auto-generated ones.
includeSegments	boolean	true (default) adds timestamped segments; false returns plain text only.
channels	array	Whole channels: links, @handles or channel IDs. Newest videos first. Use with or instead of videos.
maxVideosPerChannel	integer	1 to 500 per channel. Default 10.
publishedAfter, publishedBefore	string	Keep only channel videos uploaded in this date range (YYYY-MM-DD).
includeStats	boolean	true (default) adds likes, comment count, upload date and the channel's subscribers, handle and avatar. Views, description, channel and thumbnail are always included.
translateTo	string	Experimental. Asks YouTube for a machine translation (fr, es…). YouTube often refuses automated translation requests; you then get the original language plus a translation_note, never a failed video.
outputFormats	array	Any of srt, vtt, chunks: ready-to-use subtitle files and ~250-word RAG chunks with start/end seconds.
proxyConfiguration	object	Uses Apify Proxy by default. YouTube blocks datacenter IPs, so leave it on.
maxTotalChargeUsd	number	Hard spending cap for the run.

{
  "videos": [
    "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "jNQXAC9IVRw"
  ],
  "language": "en",
  "includeSegments": true
}

Output

One dataset item per video:

[
  {
    "video_id": "dQw4w9WgXcQ",
    "video_url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "title": "Rick Astley - Never Gonna Give You Up (Official Video)",
    "channel": "Rick Astley",
    "language": "en",
    "is_generated": false,
    "available_languages": [
      "en",
      "de",
      "es",
      "fr",
      "ja",
      "pt"
    ],
    "duration_seconds": 211.3,
    "segment_count": 61,
    "word_count": 374,
    "view_count": 1823638919,
    "like_count": 19473969,
    "comment_count": 2400000,
    "published_at": "2009-10-24",
    "channel_id": "UCuAXFkgsw1L7xaCfnd5JJOw",
    "channel_url": "https://www.youtube.com/channel/UCuAXFkgsw1L7xaCfnd5JJOw",
    "channel_handle": "@RickAstleyYT",
    "channel_subscribers": 4550000,
    "description": "The official video for “Never Gonna Give You Up” by Rick Astley...",
    "thumbnail_url": "https://i.ytimg.com/vi/dQw4w9WgXcQ/sddefault.jpg",
    "video_length_seconds": 213,
    "is_live": false,
    "text": "We're no strangers to love. You know the rules and so do I ...",
    "segments": [
      {
        "text": "We're no strangers to love",
        "start": 18.64,
        "duration": 3.24
      },
      {
        "text": "You know the rules and so do I",
        "start": 22.64,
        "duration": 4.32
      }
    ],
    "source": "youtube"
  }
]

Field	Description
video_id, video_url	The video, normalised to a watch URL
title, channel	Video title and channel name
language, is_generated	Caption language and whether YouTube auto-generated it
available_languages	Every caption language the video offers
text	The full transcript as one string
segments	[{text, start, duration}], in seconds
duration_seconds, segment_count, word_count	Size of the transcript
view_count, like_count, comment_count	Video statistics. Likes are exact; YouTube rounds the comment count on very large videos
published_at	Upload date (YYYY-MM-DD)
description, keywords, thumbnail_url	The video's description, tags and best thumbnail
channel_id, channel_url	The channel, ready to join with other data
channel_handle, channel_subscribers, channel_thumbnail_url	The channel's @handle, subscriber count (as YouTube shows it, e.g. 8,680,000) and avatar
video_length_seconds, is_live	Length of the video itself, and whether it was a live stream

A run also writes RUN_SUMMARY.json to its key-value store: videos requested, delivered, charged, and one line per failed video with the reason.

How much does it cost?

$0.005 per transcript — you pay only for videos that actually come back with a transcript.

Videos	Cost
100	$0.50
1,000	$5.00
10,000	$50.00

• Failed videos are never charged: captions disabled, private, deleted or invalid links cost $0.
• You set the ceiling: maxTotalChargeUsd (default $1) stops the run before it spends more.
• Apify's free plan includes monthly platform credit, enough for hundreds of transcripts.
• One transcript in every run of 2+ videos is free, so you can check the output on your own videos.

For comparison, the most-used transcript Actors on the Store charge $0.005–$0.01 per video.

Use cases

• LLM summarisation in a SaaS product or internal tool.
• RAG ingestion — timestamped chunks that link back to the exact second.
• AI agents that need to "watch" a video before answering.
• Search indexes over video libraries.
• Batch pipelines in n8n, Make or Airflow.

Use it from code (API)

Every Apify Actor is an API. Start a run and read the results with one call.

Python

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("om_kh/video-transcript-api").call(run_input={"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ", "jNQXAC9IVRw"], "language": "en", "includeSegments": True})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<YOUR_APIFY_TOKEN>' });
const run = await client.actor('om_kh/video-transcript-api').call({
    "videos": [
        "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
        "jNQXAC9IVRw"
    ],
    "language": "en",
    "includeSegments": true
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

One HTTP call (cURL) — runs the Actor and returns the dataset in the same response:

curl -X POST "https://api.apify.com/v2/acts/om_kh~video-transcript-api/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ", "jNQXAC9IVRw"], "language": "en", "includeSegments": true}'

Integrations: n8n, Make, Zapier, Google Sheets

• n8n — the official Apify node, operation Run Actor and get dataset, Actor om_kh/video-transcript-api.
• Make — Apify › Run an Actor, then Apify › Get Dataset Items.
• Zapier — Apify › Run Actor; map the dataset items into your next step.
• Google Sheets, webhooks, Slack — add an integration from the Integrations tab of this Actor.
• Schedules — run it every hour or day from Schedules in Apify Console.

Use it from an AI agent (MCP and x402)

Connect Claude Desktop, Cursor, VS Code or any MCP client to Apify's MCP server at https://mcp.apify.com and add this Actor, om_kh/video-transcript-api, as a tool — the agent passes the same JSON as the run input and gets the dataset back.

AI agents that pay with x402 can run it too, without an Apify account: the Actor charges only per result delivered, so an agent pays exactly for what it gets.

Tips and limits

• No captions, no transcript. This Actor reads YouTube's own captions. A video with captions turned off returns a NO_TRANSCRIPT error for that video, is not charged, and the rest of the run continues.
• Pick the language you want. If your language is missing, check available_languages in the output and run again with one of them.
• Big lists: up to 1,000 videos per run; for a whole channel or playlist use YouTube Channel Transcripts Scraper - Every Video to Text.
• Speed: videos are fetched 8 at a time; a typical video takes 1–3 seconds.

FAQ

Can I get SRT or VTT subtitle files? Yes. Set outputFormats to ["srt"], ["vtt"] or both: each result then carries a ready-to-save srt / vtt string built from the timestamped captions. No extra charge.

Is the output ready for RAG and LLMs? Yes. outputFormats: ["chunks"] adds chunks: consecutive captions merged into ~250-word pieces, each with start and end seconds, so you can embed them and cite the exact moment in the video.

Can I translate a transcript? There is an experimental translateTo option that asks YouTube for its own machine translation. YouTube often refuses automated translation requests; when it does you still get the original transcript and a translation_note (and pay only the normal price). For reliable translation, send the text or chunks to your own translation model.

Is it legal to scrape YouTube transcripts? This Actor only reads captions that YouTube shows publicly to any visitor. It does not log in and collects no personal data beyond the public channel name. You are responsible for how you use the text; check YouTube's terms and your local law, and consult a lawyer if unsure.

Does it need a YouTube API key or cookies? No. There is no key, no login and nothing that expires.

Can it transcribe videos that have no captions? No — it returns YouTube's existing captions (human or auto-generated), it does not run speech-to-text. Almost all spoken-word videos have auto-generated captions.

Which languages are supported? Every language YouTube offers captions in. Set language, or leave it empty for English first.

Can I get timestamps? Yes, includeSegments (on by default) returns every caption line with start and duration in seconds.

Am I charged for videos that fail? Never. Only transcripts that are actually delivered are charged.

Can I export to CSV or Excel? Yes, from the Storage tab of any run, or with ?format=csv on the dataset API.

Can my AI agent call it? Yes, via MCP (see above) or the plain HTTP API.

Related Actors

• YouTube Transcript Scraper - Full Text, Timestamps, Language
• YouTube Channel Transcripts Scraper - Every Video to Text
• YouTube Subtitles Scraper - Captions, Timestamps, Language
• YouTube Comments Scraper - Likes, Replies, Date, $0.40/1K
• Google News API - Headlines, Source, Date, No Login

Support

Found a bug or need a feature? Open an issue in the Issues tab of this Actor — every report gets an answer, usually within a day.