YouTube Video to Article

⚠️ ALPHA SOFTWARE — Expect breaking changes, bugs, and evolving features. Report issues →

YouTube Video to Article transforms YouTube videos into longform articles using multimodal AI. It extracts video transcripts, interprets content semantically, and synthesizes a coherent article.

Works best with educational content, talks, interviews, and explainers.

Language support: Only English is supported at the moment. Videos must have English subtitles or transcripts (manual, not auto-generated). Videos in other languages will return a NO_TRANSCRIPT error.

How it works

Three-stage pipeline:

1. Extract — Invoke YouTube Scraper to fetch video metadata and transcript
2. Interpret — Semantic analysis of transcript content (key arguments, structure, claims)
3. Synthesize — Generate a structured article grounded in the source material

Processing takes ~20–40 seconds per video, depending on transcript length.

Cost

Pay-per-event pricing:

Event	Cost
YouTube extraction (via YouTube Scraper)	Billed by that actor — see its pricing
Article synthesis	~$0.1 per article

Input

{
  "videoUrls": [
    { "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ" },
    { "url": "https://www.youtube.com/watch?v=abc123" }
  ],
  "synthesisTemplate": "article",
  "synthesisGuide": "Write in a neutral journalistic tone",
  "includeExtracted": false,
  "includeInterpretation": false
}

Parameters

Parameter	Type	Required	Default	Description
videoUrls	array	✅	—	YouTube video URLs to process. Each item is an object with url (required).
synthesisTemplate	string	No	"article"	Preset output style. Options: article, newsletter, social_thread, none.
synthesisGuide	string	No	—	Custom instructions for tone, focus, style. Appended to template instructions.
includeExtracted	boolean	No	false	Include raw extraction data (metadata + transcript) in output.
includeInterpretation	boolean	No	false	Include semantic blueprint in output.

Synthesis Templates

Templates provide base instructions for the LLM. synthesisGuide extends or refines them.

Template	Output style
article	General purpose — structured with headings, conversational voice
newsletter	Concise, scannable, hook intro, call-to-action ending
social_thread	Punchy, short paragraphs, engaging hooks — LinkedIn/Twitter style
none	No template — only custom synthesisGuide (if any) is sent

Output

Each processed video produces one dataset item:

{
  "sourceUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "title": "Generated Article Title",
  "body": "# Markdown article content...",
  "wordCount": 1250,
  "error": null,
  "errorMessage": null
}

Field	Description
sourceUrl	Original YouTube video URL
title	Generated article title (plain text)
body	Full article body in markdown
wordCount	Word count of generated article
error	Error code if failed: SCRAPE_FAILED, NO_TRANSCRIPT, INTERPRETATION_FAILED, LLM_ERROR, TIMEOUT, CONFIG_ERROR, INTERNAL_ERROR
errorMessage	Detailed error description
extracted	(only if includeExtracted: true) Raw video metadata and transcript
interpretation	(only if includeInterpretation: true) Semantic blueprint

Failed videos still produce a dataset item with error and errorMessage populated. The actor continues processing remaining URLs.

API Usage

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")
run = client.actor("actor-id").call(
    run_input={
        "videoUrls": [{"url": "https://www.youtube.com/watch?v=EXAMPLE"}],
        "synthesisTemplate": "article"
    }
)

for item in client.dataset(run["defaultDatasetId"]).list_items().items:
    print(item["title"], item["wordCount"])

import { ApifyClient } from "apify-client";

const client = new ApifyClient({ token: "YOUR_API_TOKEN" });
const run = await client
  .actor("actor-id")
  .call({
    videoUrls: [{ url: "https://www.youtube.com/watch?v=EXAMPLE" }],
    synthesisTemplate: "article",
  });

const { items } = await client.dataset(run.defaultDatasetId).listItems();
for (const item of items) {
  console.log(item.title, item.wordCount);
}

apify call actor-id \
  '{"videoUrls":[{"url":"https://www.youtube.com/watch?v=EXAMPLE"}],"synthesisTemplate":"article"}'

FAQ

What videos work best? Educational content, talks, interviews, explainers, and any video with a clear English transcript. Music videos and content without English subtitles won't produce useful results.

What if a video has no English transcript? The actor returns a NO_TRANSCRIPT error for that video and continues processing the rest. Auto-generated subtitles are not used — only manual subtitles are accepted.

Does it support other languages? Not yet. The actor currently only processes English transcripts. Support for additional languages is planned.

Can I customize the output? Yes — use synthesisTemplate for a preset style and synthesisGuide for custom instructions on tone, structure, and focus.

What if a video fails? The actor continues processing remaining URLs. Check error and errorMessage in the output for details.

Can I automate runs? Yes. Use Apify scheduling, webhooks, or the API.

Your Feedback

This is alpha software. Bug reports and feature requests help us improve:

• 🐛 Create an issue with the run ID, input parameters, and expected vs actual behavior

---

By Hikayat Labs