Real AI speech-to-text on TikTok and YouTube videos — not caption scraping. Works when captions don't exist, in any language, with per-phrase timestamps.
Most "transcript APIs" fetch the platform's existing caption track — and return nothing when there isn't one.
We run AI speech recognition on the audio itself. No captions? No problem — most short-form video has none, and that's exactly what we're built for.
Arabic transcripts get an extra AI cleanup pass. Dialects, code-switching, khaleeji — handled, with per-phrase timestamps kept intact.
Point us at a TikTok profile, YouTube channel, playlist or search query and expand it into videos — filters and sorting included, free.
Async by design — submit, then poll (or just wait). Results are cached: the same public video never costs twice.
# submit curl -X POST https://api.example.com/v1/jobs \ -H "Authorization: Bearer sk_..." \ -d '{"url": "https://www.tiktok.com/@user/video/72..."}' # -> {"id": "abc123", "status": "IN_QUEUE"} # poll curl https://api.example.com/v1/jobs/abc123 \ -H "Authorization: Bearer sk_..." # -> {"status": "COMPLETED", "output": {"text": "...", "language": "ar", # "segments": [{"start": 0.0, "end": 3.2, "text": "..."}], ...}}
Prepaid credits — top up once, we deduct per use. No subscription, and credits never expire.
| What | Price | Notes |
|---|---|---|
| Video metadata | $2 / 1,000 videos | title, description, duration, views, post date |
| AI transcription | $0.01 / started minute | real AI speech recognition — a 45s TikTok costs $0.01 |
| Translation | $10 / 1,000 videos | full transcript, flat per video — not per minute |
| Profile / channel / search expansion | Free | discovery costs nothing; the videos you fetch bill normally |
| Free tier | $0 | 1 AI transcription + 25 metadata videos each month |