Facebook Ad Library ad transcript API
Transcript of a video ad.
- Cost
- 1 creditAI transcription is billed by the started minute after 2
- Cached repeats
- 0 creditssend max_age_days, results kept 30 days
- Failed requests
- 0 credits
- Status
- Betashape may still gain fields
Ad Library videos carry no caption tracks, so this is always AI transcription of the audio, billed by the length of the audio. An ad with no video answers an unbilled error.
You send one of id or url.
Request
Send your key in the x-api-key header.
curl -G "https://api.socialscrape.io/v1/facebook/ad-library/ad/transcript" \
-H "x-api-key: $SOCIALSCRAPE_API_KEY" \
--data-urlencode 'id=352695862528588'import os
import requests
params = {
"id": "352695862528588",
}
res = requests.get(
"https://api.socialscrape.io/v1/facebook/ad-library/ad/transcript",
headers={"x-api-key": os.environ["SOCIALSCRAPE_API_KEY"]},
params=params,
timeout=60,
)
body = res.json()
if body["ok"]:
print(body["data"])
else:
print(body["error"]["code"], body["error"]["message"])const url = new URL("https://api.socialscrape.io/v1/facebook/ad-library/ad/transcript");
url.search = new URLSearchParams({
id: "352695862528588",
}).toString();
const res = await fetch(url, {
headers: { "x-api-key": process.env.SOCIALSCRAPE_API_KEY! },
});
const body = await res.json();
if (body.ok) {
console.log(body.data);
} else {
console.error(body.error.code, body.error.message);
}{
"nodes": [
{
"name": "SocialScrape Facebook Ad Library ad transcript",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
0,
0
],
"parameters": {
"method": "GET",
"url": "https://api.socialscrape.io/v1/facebook/ad-library/ad/transcript",
"sendQuery": true,
"queryParameters": {
"parameters": [
{
"name": "id",
"value": "352695862528588"
}
]
},
"sendHeaders": true,
"headerParameters": {
"parameters": [
{
"name": "x-api-key",
"value": "={{ $env.SOCIALSCRAPE_API_KEY }}"
}
]
}
}
}
],
"connections": {}
}Parameters
| Name | Type | Required | Description |
|---|---|---|---|
id | string | one of | Ad id in the library. Also accepted as ad_id, adId. |
url | url | one of | Ad Library URL (facebook.com/ads/library/?id=...). |
Every request also accepts trim, raw and max_age_days.
Response
Sample from a recorded response, shortened. Values are public data, not yours.
{
"ok": true,
"credits_charged": 1,
"credits_remaining": 4999,
"cached": false,
"cached_at": null,
"request_id": "req_63eae290249d0c81df4c4ff3",
"data": {
"object": "transcript",
"platform": "facebook",
"video_id": "352695862528588",
"url": "https://www.facebook.com/ads/library/?id=352695862528588",
"language": "en",
"source": "ai",
"duration_seconds": 30,
"text": "canned transcript for the fixture, not the ad's real audio",
"segments": [
{
"start": 0,
"end": 4,
"text": "canned transcript for the fixture,"
},
{
"start": 4,
"end": 30,
"text": "not the ad's real audio"
}
],
"available_languages": null,
"extra": {
"media_seconds": 30
}
},
"cursor": null,
"has_more": null
}Fields
One Transcript in data.
A field the platform does not give is null, never made up. Extra platform details are in extra.
| Field | Type | Always | Notes |
|---|---|---|---|
object | "transcript" | ||
platform | Platform | yes | |
video_id | string or null | ||
url | string or null | ||
language | string or null | BCP 47, "en" | |
source | "captions" or "auto_captions" or "ai" | yes | |
duration_seconds | number or null | ||
text | string | yes | |
segments | TranscriptSegment[] or null | ||
available_languages | string[] or null | ||
extra | object or null |
Full field list in the reference for this endpoint and the data models. Errors are listed under Errors.
Questions
Which errors can it return?
invalid_params (do not retry), unsupported_url (do not retry), not_found (do not retry), transcript_unavailable (do not retry). Account and platform errors from the errors list can also happen. A failure always costs 0 credits.
Is there an SDK call for it?
The preview Python client has client.facebook.ad_library_ad_transcript(...) and the TypeScript client client.facebook.adLibraryAdTranscript(). Both return the same envelope as the HTTP API. The packages are not published yet; plain HTTP works today in any language.
Can an AI agent call it?
Yes. The MCP server reaches every built endpoint, with the same prices.