All guides

November 15, 2026 · 6 min read

YouTube Data API captions.download 403 Forbidden: Why It Happens and How to Fix It

In short

captions.download returns 403 forbidden unless your OAuth user can edit the video. Fix it for your own videos, or get any public transcript in one call.

The YouTube Data API returns 403 forbidden from captions.download because that endpoint only works when an OAuth-authorized user has permission to edit the video, so it fails for any video you don't own or manage, and it never works with an API key alone. If the video is yours, fix the OAuth scope and the account; if it isn't, no setting will make captions.download work, and you need a transcript API such as TranscriptYT that reads the public caption track in one request.

What the captions.download 403 forbidden error means

Google's reference for captions.download says the request must be authorized with the youtube.force-ssl or youtubepartner scope, and that the authorized user must have permission to edit the video. The 403 with reason forbidden means the permissions behind the request are not sufficient to download that caption track.

That is a permission check, not a quota or rate-limit problem. Retrying, waiting, or raising your quota changes nothing. The typical response looks like this:

GET https://www.googleapis.com/youtube/v3/captions/CAPTION_ID?tfmt=srt
Authorization: Bearer ya29....

HTTP/1.1 403 Forbidden
{
  "error": {
    "code": 403,
    "message": "The permissions associated with the request are not sufficient to download the caption track. ...",
    "errors": [{ "domain": "youtube.caption", "reason": "forbidden" }]
  }
}

Quick check: which kind of 403 do you have?

Before changing code, find out whether the problem is your setup or the video itself. These three questions settle it in a minute:

  • Is your request sending an OAuth access token (Authorization: Bearer ya29...) rather than only key=YOUR_API_KEY? If not, you are not authorized at all.
  • Open YouTube Studio while signed in as the same Google account. Does the video appear in that channel's content list? If not, the account cannot edit it and the 403 is permanent.
  • Did the token you are using get granted youtube.force-ssl? Inspect it at https://oauth2.googleapis.com/tokeninfo?access_token=... and read the scope field.

Why captions.list works but captions.download returns 403

This is the part that confuses most people. captions.list accepts a plain API key and happily returns caption track IDs for any public video, so it looks as if the next step should work too. But captions.list only returns metadata about each track (ID, language, name, whether it is auto-generated). The response does not contain the caption text.

The text only comes from captions.download, and that endpoint is gated on edit rights. So you can list the tracks of someone else's video, but you cannot download them. A call with only an API key and no OAuth token usually fails with 401 instead of 403; a valid OAuth token from an account that does not manage the video gets the 403.

Quota makes the dead end more expensive: captions.list costs 50 units and captions.download costs 200 units, out of a default daily allocation of 10,000 units per project. Even when it works, that is at most 50 downloads a day on default quota before you count the list calls.

How to fix the 403 for videos you own or manage

If the video belongs to your channel, or to a channel where your Google account has an editing role, the 403 is fixable. Go through these checks in order:

# pip install google-api-python-client google-auth-oauthlib
from google_auth_oauthlib.flow import InstalledAppFlow
from googleapiclient.discovery import build

SCOPES = ["https://www.googleapis.com/auth/youtube.force-ssl"]
flow = InstalledAppFlow.from_client_secrets_file("client_secret.json", SCOPES)
creds = flow.run_local_server(port=0)  # sign in as the channel owner

yt = build("youtube", "v3", credentials=creds)
tracks = yt.captions().list(part="snippet", videoId="YOUR_VIDEO_ID").execute()
caption_id = tracks["items"][0]["id"]

srt = yt.captions().download(id=caption_id, tfmt="srt").execute()
open("captions.srt", "wb").write(srt)
  • Use OAuth 2.0, not an API key. captions.download needs a user access token in the Authorization header.
  • Request the right scope: https://www.googleapis.com/auth/youtube.force-ssl (or youtubepartner for content owners). The read-only youtube.readonly scope is not enough.
  • Sign in with the account that owns the channel. A common mistake is authorizing with a personal Google account while the channel lives on a Brand Account; pick the channel in the account chooser during consent.
  • If you added the scope after the first consent, revoke the app at myaccount.google.com/permissions and sign in again, or delete your stored token file, so the new scope is actually granted.
  • Check the caption ID comes from captions.list for that same video, and that the track is not a draft you lack access to.
  • Pass tfmt (srt, vtt, sbv, scc, or ttml) if you want a subtitle file rather than the original upload format.

Why you can't fix it for other people's videos

If the video is on someone else's channel, every one of those fixes still ends in 403. Edit permission belongs to the channel owner and the managers they add, and there is no scope, API key, or quota increase that grants it to a third party. Your options with the official API are to ask the owner to add you as a manager or to have them authorize your app, which does not scale past a handful of channels.

Yet the same captions are visible to anyone who presses CC on the watch page. That gap between what viewers can read and what the Data API will give you is why public-transcript APIs exist. For a longer comparison of the two models, see YouTube Transcript API vs. YouTube Data API.

Get any public video's captions in one call

The TranscriptYT API skips the list-then-download dance and the OAuth flow. You send a video URL or ID with an API key and get the transcript back: no caption IDs, no channel ownership, and no Google quota units. The request reads the public caption track (creator-uploaded or auto-generated) that YouTube shows viewers.

Output can be json (default, with timestamped segments, title, channel, and the list of available languages), text, srt, vtt, csv, or md. Pick a track with language, translate with translate_to into any of 150+ languages, or choose manual or auto captions with captionType.

curl "https://transcript-yt.com/v1/transcript?url=https://youtu.be/5f3Pn-N6yLc&format=srt" \
  -H "Authorization: Bearer ts_live_xxxx" -o captions.srt

# Python
import os, requests
res = requests.get(
    "https://transcript-yt.com/v1/transcript",
    params={"url": "5f3Pn-N6yLc", "format": "json"},
    headers={"Authorization": f"Bearer {os.environ['TRANSCRIPTYT_KEY']}"},
)
res.raise_for_status()
data = res.json()
print(data["language"], data["autoGenerated"], data["segments"][:2])

Mapping captions.list and captions.download to TranscriptYT endpoints

If you are replacing Data API code, the calls map almost one to one:

  • captions.list (50 units) becomes GET /v1/languages, which lists a video's caption tracks plus every translation language and costs no credits.
  • captions.download (200 units, owner only) becomes GET /v1/transcript, 1 credit per successful transcript for any public video. Repeat requests served from cache also cost 1 credit.
  • tfmt=srt or tfmt=vtt becomes format=srt or format=vtt; there is also text, csv, md, and JSON with segments.
  • tlang (machine translation) becomes translate_to, which keeps each segment's timing.
  • Many videos at once: POST /v1/transcripts takes a list of URLs or a whole playlist or channel, from 10 per call on Free up to 500 on Scale.
  • Failed requests are never billed, and errors use one envelope with codes such as NO_CAPTIONS, VIDEO_UNAVAILABLE, and RATE_LIMITED.

Pricing, limits, and videos without captions

The Free plan gives 100 credits once, 10 requests a minute, and captions only, enough to test your migration. Paid plans start with Micro at $3 a month for 1,000 credits, with Starter at $9 for 4,000, Growth at $29 for 15,000, and Scale at $79 for 50,000, all at 200 requests a minute.

Some videos have no caption track at all, which captions.download could not help with either. On paid plans and credit packs, mode=auto (the default) transcribes the audio with AI speech-to-text when captions are missing, billed at 1 credit per started minute; mode=captions never uses AI. The no-captions guide covers that case in more detail.

If you only need one or two files and no code, paste the link into the free YouTube SRT download tool or the YouTube transcript tool.

Common questions

Why does captions.download return 403 forbidden?

Because the authorized user does not have permission to edit the video. captions.download only works with an OAuth token (youtube.force-ssl or youtubepartner scope) from the channel owner or a manager of that channel.

Can I download captions from any YouTube video with the Data API?

No. captions.list works for public videos, but it returns only track metadata, and captions.download refuses videos you can't edit. For other people's public videos, use a transcript API that reads the public caption track.

Does captions.download work with an API key?

No. It requires an OAuth 2.0 user access token, and an API key alone is rejected. captions.list, by contrast, accepts a plain API key for public videos but returns no caption text.

How much quota does captions.download use?

200 units per call, plus 50 for the captions.list call that finds the caption ID. With the default 10,000 units a day that caps you at about 40 list-plus-download pairs.

Is there a YouTube captions API that doesn't need OAuth?

Yes. TranscriptYT's GET /v1/transcript takes a video URL and an API key and returns the public captions as JSON, text, SRT, VTT, CSV, or Markdown for 1 credit, with 100 free credits to start.

Try the YouTube Transcript API

Get transcripts as JSON, text, SRT, or VTT. 100 free credits, no card required.

More guides