YouTube Transcript Fast

What it does

HTTP-only YouTube transcript extraction — no Playwright/browser. For each video: fetches the watch page, pulls captions.playerCaptionsTracklistRenderer.captionTracks out of the inline ytInitialPlayerResponse, picks a track (manual caption in the requested language, else auto-generated in that language, else any manual track, else any track), fetches that track with &fmt=json3, and assembles the full transcript text plus timestamped segments.

Input

Output example

One row per video in the default dataset:

{
  "videoId": "dQw4w9WgXcQ",
  "title": "Rick Astley - Never Gonna Give You Up",
  "channel": "Rick Astley",
  "duration": 213,
  "language": "en",
  "isAutoCaption": false,
  "transcript": "We're no strangers to love ...",
  "segments": [ { "start": 0.5, "dur": 2.1, "text": "We're no strangers to love" } ]
}

A video with no caption tracks pushes:

{ "videoId": "...", "title": "...", "transcript": null, "reason": "no-captions" }

No charge event fires for that row.

Pricing

Pay-per-event. One transcript-scraped event fires per successfully pushed transcript row, via Actor.charge({ eventName: 'transcript-scraped' }). Configure in the Apify Console: $0.0008 per transcript-scraped event (docs/CANDIDATES.md entry 2 — 12x under the 3.72-rated market leader's $0.01/item). No-caption rows are never charged.

Known gap — BLOCKED from this Mac, not worked around

The watch-page fetch and ytInitialPlayerResponse extraction work (200 OK, tracks parsed correctly, verified against dQw4w9WgXcQ and others). The second call — fetching the caption track itself (https://www.youtube.com/api/timedtext?...&fmt=json3) — consistently comes back HTTP 200 with an empty body (content-length: 0,

server: video-timedtext

Limits

Free-plan compute only (policy/RULES.md rule 4). No proxy configured (rule 1 — target needs no residential proxy in the common case; see gap above for the one endpoint that may need it). No personal data beyond public channel/title/video metadata already documented in docs/CANDIDATES.md.