Is my Short silent? Check has_audio before detach or captions

Run video inspect with frames: false and read probe.has_audio. If it is false, skip audio detach and speech captions and use caption cues instead. The Sume docs say a frames: false inspect is enough to learn this.

Which refusals you avoid

Silent b-roll Shorts are common. Sending one to the wrong tool wastes a request, and each tool refuses with a specific code.

The check

Import the clip to media.sume.com first; inspect does not fetch from the open internet. Then send the clip with no frames and no transcript. The result carries a probe with duration, size, fps, codecs and has_audio.

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-broll-01" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/broll.mp4",
    "frames": false
  }'

Branch on the answer

If has_audio is true, detach it ($0.01) or caption it from speech ($0.20 flat for clips up to 60 seconds). If false, send authored text with cues and your own start and end times. Cues burn exactly your text, so no speech recognition runs.

Cost

Inspect without a transcript is read-only, and the Sume docs describe it as free apart from worker compute. Only transcribe: true bills the speech-to-text rate of $0.01 per audio minute.

Series habit

Keep the probe result with the episode notes. For a series, a quick run on every episode catches a file exported without its audio track before you promise captions.