Facebook Transcript Scraper

Facebook Transcript Scraper — videos and reels to text

The Facebook Transcript Scraper turns public Facebook videos and reels into text. It uses the caption track Facebook publishes when there is one, and otherwise transcribes the audio with a built-in Whisper speech model.

It runs on Apify, needs no Facebook login and no cookies, and exports to JSON, CSV, Excel or an API call. Every session key it needs (tokens, query ids, cursors) is read from Facebook on each run, so it keeps working when Facebook ships new code.

What the Facebook Transcript Scraper does

• Captions first: exact text with timestamps when Facebook has them.
• Speech recognition second: Whisper model built into the Actor.
• Returns full transcript, timed segments, language and the source used.
• Marks music-only videos with hasSpeech=false.

Who uses the Facebook Transcript Scraper

• Creators repurposing videos into articles.
• Analysts studying what pages say on video.
• Accessibility teams needing text versions.

How the Facebook Transcript Scraper works

Facebook serves public pages, videos and listings to logged-out visitors and embeds the first page of data in the HTML.

The Actor loads that page like a browser, reads the session values (LSD token, revision, query ids) Facebook writes into it, and requests the following pages from Facebook's GraphQL endpoint with the returned cursor.

Query ids that a page does not preload are read from the JavaScript bundles the page lists, and the variable names of each query come from its own module, so the request always matches Facebook's current code.

Nothing is guessed: if Facebook stops publishing a value the Facebook Transcript Scraper needs, the run log names it and the item is skipped instead of being filled with a stale value.

Input of the Facebook Transcript Scraper

Field	Type	Default	What it does
videoUrls	array	["https://www.facebook.com/reel/1549207066583030"]	Facebook video, reel, watch or share links.
speechRecognition	boolean	True	Transcribes the audio when Facebook publishes no caption track.
language	string	—	Two-letter code such as en, es, pt. Empty = detect automatically.
maxTotalItems	integer	0	Stops the whole run after this many dataset rows.
proxyConfiguration	object	{"useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"]}	Facebook limits logged-out visitors per IP address. Residential proxies (the default) give the most reliable results; the Actor switches to a fresh IP when Face

Example input for the Facebook Transcript Scraper:

{
  "videoUrls": [
    "https://www.facebook.com/reel/1549207066583030"
  ],
  "speechRecognition": true
}

Output of the Facebook Transcript Scraper

Each dataset row is one transcribed video. These are the fields the Facebook Transcript Scraper produced in its test runs:

Field	Meaning
captionLocales	captionLocales
durationSec	Length in seconds
hasSpeech	False when the audio has no speech (music only)
id	Platform id of the item
inputUrl	inputUrl
language	Detected language
languageProbability	languageProbability
ownerName	Page or creator name
playCount	Video plays
publishedAt	Publication time (UTC, ISO 8601)
scrapedAt	When this row was read
segments	Transcript segments with start/end seconds
text	Post or video text
transcript	Full transcript text
transcriptSource	transcriptSource
url	Public link of the item

Fields that Facebook does not publish for an item are left empty rather than guessed. Download the dataset as JSON, CSV, Excel, XML or HTML, or read it through the Apify API.

Example row from a Facebook Transcript Scraper test run

A real row from the local test run (long values shortened):

{
  "id": "1665351285019129",
  "url": "https://www.facebook.com/reel/1665351285019129/",
  "text": "This October, you can look forward to these sights in the night sky:\n- A good look at S...",
  "ownerName": null,
  "publishedAt": "2026-10-02T14:11:36.000Z",
  "durationSec": 162.37,
  "playCount": 34967,
  "inputUrl": "https://www.facebook.com/NASA/videos/whats-up-for-october-2026/1665351285019129/",
  "transcriptSource": "speech-recognition",
  "language": "en",
  "languageProbability": 0.998,
  "transcript": "Saturn takes center stage, a meteor shower lights up the sky, and the moon visits the s...",
  "segments": [
    {
      "start": 0.0,
      "end": 6.48,
      "text": "Saturn takes center stage, a meteor shower lights up the sky, and the moon visits the s..."
    },
    {
      "start": 6.48,
      "end": 7.48,
      "text": "sisters."
    },
    {
      "start": 7.48,
      "end": 10.16,
      "text": "That's what's up for October."
    },
    "..."
  ],
  "hasSpeech": true,
  "captionLocales": null,
  "scrapedAt": "2026-10-02T18:56:33.000Z"
}

Data quality of the Facebook Transcript Scraper

• Every value in a Facebook Transcript Scraper row is read from Facebook during your run; nothing is filled from a cache or an old snapshot.
• Rows are de-duplicated by id inside a run, so the same transcribed video never appears twice in one dataset.
• Counts are numbers you can sort and sum (not "1.2K" text) wherever Facebook publishes the exact number.
• Dates are UTC in ISO 8601 format, which Excel, Google Sheets, Python and SQL all read directly.

Facebook Transcript Scraper test results

These numbers come from live test runs of the Facebook Transcript Scraper on 2026-10-03 (home connection, no proxy). Runs on the Apify platform use residential proxies and can be faster or slower.

Test	Rows	Time
reel without captions -> speech recognition	1	7 s
captions disabled + no ASR	1	3 s

Facebook Transcript Scraper limits, stated plainly

• Measured locally: a 162-second NASA video transcribed into 35 segments in 24 seconds.
• Speech recognition can mishear names and numbers.
• Music-only reels return an empty transcript (measured on a 46 s reel).

How to run the Facebook Transcript Scraper

1. Open the Facebook Transcript Scraper in Apify Console and fill in the input fields described above.
2. Keep the default residential proxy unless you have a reason to change it; Facebook limits logged-out visitors per IP address.
3. Start the run, watch the log, and download the dataset when the status says Done.
4. To automate it, call the Facebook Transcript Scraper from the Apify API, schedule it, or connect it to Make, Zapier, n8n or Google Sheets through an integration or webhook.

Python example with the Apify API client:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run_input = {
    "videoUrls": [
        "https://www.facebook.com/reel/1549207066583030"
    ],
    "speechRecognition": True
}
run = client.actor("neuro-scraper/facebook-transcript-scraper").call(run_input=run_input)
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

Facebook Transcript Scraper pricing

The Facebook Transcript Scraper uses pay-per-event pricing: you pay per dataset row (event "result"), plus the Apify platform usage of your run. Free-plan users get up to 30 rows per run, which is enough to test the Facebook Transcript Scraper on real data.

Set a maximum cost per run in Apify and the Facebook Transcript Scraper stops cleanly when it is reached.

Tips for large runs of the Facebook Transcript Scraper

• Start with a small limit to check the Facebook Transcript Scraper output, then raise it; the first rows arrive within seconds.
• Split very large jobs into several inputs and run them in parallel tasks; each run gets its own proxy sessions.
• Set a maximum cost per run in Apify; the Facebook Transcript Scraper stops cleanly when the budget is reached and keeps every row it already saved.
• Schedule repeat runs instead of one huge run when you track the same targets over time.
• Keep the residential proxy: datacenter IPs are refused by Facebook far more often for logged-out visitors.

Facebook Transcript Scraper compared with copying by hand

	Facebook Transcript Scraper	Manual work
Speed	Seconds per page of results	Minutes per item
Fields	Every public field, same order every time	Whatever you copy
Format	JSON, CSV, Excel, API	Spreadsheet you build yourself
Repeatable	Schedule it daily	Start again each time
Errors	Logged per item	Unnoticed typos

Using the Facebook Transcript Scraper in a workflow

• Send every new row to Google Sheets or Airtable with the Apify integration and keep a live table.
• Trigger a Make, Zapier or n8n scenario with a webhook when the run finishes.
• Load the JSON export into Python or BigQuery for analysis and dashboards.
• Use the Apify API: start the Facebook Transcript Scraper with a POST request and read the dataset items endpoint when the run finishes.
• Chain it: pass the Facebook Transcript Scraper output to another Apify Actor (for example a downloader or a transcript Actor) in an Apify task.
• Export straight to Excel from the dataset tab when you only need a one-off report.

Troubleshooting the Facebook Transcript Scraper

• The run ends with no rows: open the log. The Facebook Transcript Scraper writes one line per target saying what Facebook returned, for example a login wall or a missing item.
• Fewer rows than expected: the target may have fewer public items, or Facebook may stop serving more pages to logged-out visitors. The log names the reason.
• Slow run: large limits mean many pages; the Facebook Transcript Scraper waits briefly between pages so Facebook keeps answering. Split big jobs into several runs.
• Proxy errors: keep the RESIDENTIAL group. The Facebook Transcript Scraper already switches to a fresh IP when one is blocked.
• Free plan stops at 30 rows: that is the free-plan cap of the Facebook Transcript Scraper; any paid Apify plan removes it.

What changed in this version of the Facebook Transcript Scraper

• Rebuilt in October 2026 on Facebook's logged-out web endpoints; every token, query id and cursor is read at run time.
• Pay-per-event pricing (one event per row) replaces the old monthly rental; free-plan users can test with 30 rows per run.
• Residential proxy by default, with automatic switch to a fresh IP when Facebook blocks one.

Is it legal to use the Facebook Transcript Scraper?

The Facebook Transcript Scraper reads only data that Facebook shows to any logged-out visitor. It never logs in and never touches private content.

Personal data may still be in public posts, so check GDPR and the other laws that apply to you before you store or reuse it, and respect Facebook's terms.

Related scrapers

Actor	Link
Facebook Posts Scraper	neuro-scraper/facebook-posts-scarper
Facebook Reels Scraper	neuro-scraper/facebook-reels-scraper
Facebook Comments Scraper	neuro-scraper/facebook-comments-scraper
Facebook Video Details Scraper	neuro-scraper/facebook-video-details-scraper
Facebook Video Downloader	neuro-scraper/facebook-video-downloader
Facebook Reels Downloader	neuro-scraper/facebook-reels-downloader
Facebook Video Scraper	neuro-scraper/facebook-video-scraper-pro
Facebook Photos Scraper	neuro-scraper/facebook-photo-posts-scraper
Facebook Reels Scraper Advanced	neuro-scraper/facebook-reels-scraper-advanced
Facebook Ads Library Scraper	neuro-scraper/facebook-image-photos-downloader
Facebook Pages Scraper	neuro-scraper/facebook-reels-downloader-and-facebook-video-downloader
Facebook Marketplace Scraper	neuro-scraper/facebook-video-posts-scraper

Leave a review

If the Facebook Transcript Scraper saved you time, please leave a star rating and a short review on its Apify page. Reviews help other users find the Facebook Transcript Scraper and tell us what to improve next.

Frequently asked questions

Where does the Facebook Transcript Scraper transcript come from?

transcriptSource says it: "facebook-captions" when Facebook publishes captions, "speech-recognition" when the Facebook Transcript Scraper transcribed the audio.

Does the Facebook Transcript Scraper need my Facebook login?

No. The Facebook Transcript Scraper works logged out and never asks for your password or cookies. It reads only what Facebook shows to any visitor.

Do I need my own proxy for the Facebook Transcript Scraper?

No. The Facebook Transcript Scraper uses Apify residential proxies by default and switches to a fresh IP when Facebook blocks one. You can choose another proxy in the input.

What formats can I export from the Facebook Transcript Scraper?

JSON, CSV, Excel, XML, HTML and RSS from the dataset tab, or any format through the Apify API, webhooks and integrations.

How fast is the Facebook Transcript Scraper?

In the test runs above the Facebook Transcript Scraper took between 3 and 7 seconds per run. Speed depends on how many items you ask for and on the proxy.

Why did the Facebook Transcript Scraper return fewer rows than I asked for?

The Facebook Transcript Scraper stops when Facebook has nothing more to show logged out, when your limit is reached, or when the free-plan cap of 30 rows applies. The run log always says which one happened.

Can I schedule the Facebook Transcript Scraper?

Yes. Create a schedule in Apify Console to run the Facebook Transcript Scraper daily or hourly and send the results to Google Sheets, Make, Zapier or n8n through an integration.

Can the Facebook Transcript Scraper read private accounts or private groups?

No. The Facebook Transcript Scraper reads public data only. Private content needs a login, and the Facebook Transcript Scraper never logs in.

What happens when Facebook changes its website?

The Facebook Transcript Scraper reads every key it needs from Facebook at run time, so routine changes are picked up automatically. If something can no longer be read, the log says so instead of returning wrong data.

Can I use the Facebook Transcript Scraper with Make, Zapier or n8n?

Yes. Add the Facebook Transcript Scraper as an Apify step in Make, Zapier or n8n, or send a webhook when the run finishes and fetch the dataset items from the Apify API.

Does the Facebook Transcript Scraper work in every country?

The Facebook Transcript Scraper reads Facebook through residential proxies that rotate across countries. Content that Facebook restricts by country follows those rules.

How do I collect only new items with the Facebook Transcript Scraper?

Schedule the Facebook Transcript Scraper and keep the item ids from earlier runs. Every row has a stable id, so a simple de-duplication against your previous export keeps only the new ones.

Can the Facebook Transcript Scraper process many targets in one run?

Yes. Put as many targets as you like in the input; the Facebook Transcript Scraper handles them one after another and a failing target never stops the others.