Get a transcript of an AI video before TikTok Shop review

To get a transcript of an AI video before you post it, call POST /v1/video-inspect with transcribe: true on a media.sume.com clip. The transcript is billed at Sume's public STT rate of $0.01 per audio minute, on top of the inspect's compute, and it gives you the exact words so you can compare them to your listing.

That comparison matters for TikTok Shop. The US page bans exaggerated product effects and fabricated expert personas, and the UK page bans fabricating product capabilities. A voice-over that a model improvised can say something your script never did.

How to run it

Video inspect reads one clip that the workspace already owns on media.sume.com; it will not fetch from the open internet, so import first with POST /v1/media-imports. The default mode is sync, which waits up to 30 seconds for a 200 and otherwise returns 202, and you poll the job.

Send frames: false if you want no stills. Send duration_seconds (up to 600) as a hint so the reserve matches the clip; without it Sume reserves one minute. language_code is an STT hint, and with segmentation.mode: "sentence" you also get sentence segments.

curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: inspect-transcript-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4",
    "frames": false,
    "transcribe": true,
    "language_code": "en",
    "segmentation": {"mode": "sentence"}
  }'

What to do with the text

Read the transcript against the claims on your product page. Strike any line that promises a result, a medical benefit, or a comparison you cannot support. Then fix the clip: re-generate the voice, or trim the line out with Audio detach and Timeline if you only need to drop a sentence.

Keep the transcript with the post record. Both Shop pages ask for disclosure when AI plays a significant role in the content, and a stored transcript shows what the final video actually said.

Edge cases

A silent clip fails with inspect_source_has_no_audio. If you send language_code, segmentation, or duration_seconds without transcribe: true, you get 400 video_inspect_transcribe_required. Inspect also bills its own container seconds, so the total is more than the STT line; the docs do not give a fixed number for that part.