Facebook AI Transcript Extractor

Turn anonymously accessible Facebook videos and Reels into searchable transcripts, timestamped segments, SRT subtitles, and WebVTT captions. Use this Facebook AI transcript extractor for research, accessibility drafts, content analysis, or downstream AI workflows; private and login-gated media are not supported.

Quick start

1. Paste a public Facebook Reel or video URL into Facebook video or Reel URLs.
2. Start with one short video. Leave language empty unless you know the source language.
3. Run the Actor, then open the dataset to export JSON, CSV, or Excel. Save srtSubtitles or vttSubtitles as a subtitle file.

Use the full facebook.com/reel/ID or facebook.com/PAGE/videos/ID URL, not a fb.watch, share, or Watch-page link.

Example input:

{
  "videoUrls": [
    "https://www.facebook.com/cnn/videos/10155529876156509/"
  ],
  "maxVideoDurationSeconds": 900,
  "language": "en"
}

Output example

Illustrative shape only; transcript text and timing below are not a captured quotation. Optional metadata may be absent.

{
  "status": "succeeded",
  "sourceUrl": "https://www.facebook.com/cnn/videos/10155529876156509/",
  "canonicalUrl": "https://www.facebook.com/cnn/videos/10155529876156509/",
  "videoId": "10155529876156509",
  "transcript": "Timestamped transcript text...",
  "language": "en",
  "segments": [{ "text": "Timestamped transcript text...", "startSeconds": 0, "endSeconds": 4.2 }],
  "srtSubtitles": "1\n00:00:00,000 --> 00:00:04,200\nTimestamped transcript text...",
  "vttSubtitles": "WEBVTT\n\n00:00:00.000 --> 00:00:04.200\nTimestamped transcript text..."
}

Input settings

Setting / API key	Type	Description
Facebook video or Reel URLs / videoUrls	array	Required: 1–20 public video URLs, as strings or { "url": "..." } objects.
Maximum video duration / maxVideoDurationSeconds	integer	10–7200 seconds; omitted input uses 900 seconds.
Language hint / language	string	Optional source language, such as en or es; not a translation target. See field help for selection behavior.

Output fields

Each successful video produces one dataset row.

Field	Description
status	Processing status.
sourceUrl, canonicalUrl	Submitted URL and the source's video-page URL when available.
videoId, videoTitle, description	Public video identity and text metadata.
durationSeconds	Video duration.
uploaderName, uploaderId, publishedAt	Public uploader and publication metadata when available.
thumbnailUrl	Public preview image URL when available.
mediaFormat	Selected public media format when available.
transcript, language	Full transcript and selected/requested language; may be unknown.
segments	Timestamped objects with text, startSeconds, and endSeconds.
srtSubtitles, vttSubtitles	Ready-to-save subtitle documents.

Input recipes

• One short Reel: submit one URL and set maxVideoDurationSeconds to 120.
• Small research batch: submit up to 20 public URLs; start small because long videos may leave later targets unfinished.
• Known source language: set language to the spoken language, not the language you want to translate into.

Pricing

The Actor uses pay per event: start once per run after input validation, and apify-default-dataset-item (Video processed) for each transcript saved to the dataset. Failed targets do not produce a saved row or incur the per-transcript event; a run with no saved transcripts can still incur the start event. See the live Pricing tab for current rates and plan-based volume discounts.

Tips and limits

• Only public, directly addressable Facebook videos and Reels are supported; login-only, deleted, private, geo-restricted, or rights-restricted media may fail.
• Clear speech produces better results than music, overlapping speakers, or loud background noise.
• A batch preserves successful rows when another URL fails. RUN_SUMMARY and STATE in the key-value store describe failures and unfinished targets.
• Repeated URLs are processed once, in input order. There is no search, sorting, speaker labeling, or translation setting.
• Missing captions or unintelligible audio can prevent a transcript. A batch with no saved transcripts fails; do not treat every inaccessible URL as private or deleted.
• Timestamps and speech recognition are drafts: review important names, numbers, and quotations.

Who is it for?

• Researchers and journalists who need searchable quotations and timestamps from public videos.
• Accessibility teams preparing subtitle drafts for authorized content.
• Content and social teams indexing spoken material across a small URL batch.
• Developers and AI builders feeding structured transcripts into search, summarization, or retrieval pipelines.

API usage

Start a run with cURL:

curl -X POST "https://api.apify.com/v2/acts/fetch_cat~facebook-ai-transcript-extractor/runs" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"videoUrls":["https://www.facebook.com/cnn/videos/10155529876156509/"]}'

JavaScript with apify-client:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('fetch_cat/facebook-ai-transcript-extractor').call({
  videoUrls: ['https://www.facebook.com/cnn/videos/10155529876156509/'],
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

Python with apify-client:

import os
from apify_client import ApifyClient
client = ApifyClient(os.environ['APIFY_TOKEN'])
run = client.actor('fetch_cat/facebook-ai-transcript-extractor').call(run_input={
    'videoUrls': ['https://www.facebook.com/cnn/videos/10155529876156509/'],
})
items = client.dataset(run['defaultDatasetId']).list_items().items
print(items)

MCP and AI agents

Use the official Apify MCP server, not a server built into this Actor. The focused URL selects this Actor; the general server can discover other Actors. Follow Apify's MCP setup guide for authentication. Start with one public URL and read transcript and segments from the returned dataset.

$claude mcp add --transport http apify https://mcp.apify.com/?tools=fetch_cat/facebook-ai-transcript-extractor

Or add the server to an MCP client configuration:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=fetch_cat/facebook-ai-transcript-extractor"
    }
  }
}

Example prompts:

• “Transcribe this public Facebook Reel and return the timestamps for every mention of the product name.”
• “Create an SRT subtitle draft for these public Facebook video URLs.”
• “Compare the themes in these transcripts and cite the relevant time ranges.”

Keep URL validation and human review in sensitive workflows.

FAQ

Does it work without a Facebook login?

It processes public URLs that Facebook exposes anonymously. It does not bypass login, privacy, or access restrictions.

Can I export Facebook video transcripts as SRT or WebVTT?

Yes. Save the srtSubtitles or vttSubtitles field from each dataset item as an .srt or .vtt file.

Why did a public URL fail?

The media may have no audible speech, may be temporarily unavailable, or may be restricted by location, rights, age, or Facebook delivery changes. Review the run log and RUN_SUMMARY.

Can I run Facebook video transcription through an API or MCP client?

Yes. Use the Apify API, official clients, webhooks, integrations, or the MCP endpoint above.

Is this an alternative to manual transcription?

It automates first-pass transcription for small URL batches. Review important names, numbers, and low-quality audio before publication or compliance use.

Related Actors

Continue a Facebook content workflow with:

• Facebook Pages Scraper
• Facebook Ads Library Scraper
• Facebook Comments Scraper
• Facebook Events Scraper
• Facebook Reels Scraper

Support and feedback

For custom additions or integrations, email hello@fetchcat.dev.

Found an error or have a feature request? Open an issue with the run ID, a non-sensitive input example and what you expected. Do not include credentials or private media.

Made with care for curious people, AI agents and the possibilities in data. ♥