Audio detach range without end: take the rest of a track

In Sume audio detach, range is { start, end? } in seconds. If you leave out end, the range is open-ended and runs to the end of the track. The audio detach docs, read 2026-10-05, say this directly. Omit range altogether and you get the whole track.

Open-ended is handy when you know where the speech starts, for example after a cold open, and you want everything after it.

The caps still apply

Output is documented as capped at 900 seconds, and the source at 1800. The worker checks the 900 second cap when you send both start and end (a range over 900 seconds fails with audio_detach_range_empty, the same code used for an end before the start) and when you omit range on a source longer than 900 seconds (output_duration_exceeded). I did not find a check that rejects a start-only range whose remaining length is over 900 seconds, so do not rely on one: size the range yourself and confirm duration_seconds in the result.

Other range errors

• detach_start_past_source: start is beyond the probed duration.
• detach_source_has_no_audio: the video has no audio stream.
• source_duration_exceeded: the source is over 1800 seconds.

Request

curl -X POST https://api.sume.com/v1/audio-detach \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: detach-tail-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4",
    "range": {"start": 600},
    "channels": "mono",
    "sample_rate": 16000
  }'

The result reports duration_seconds, source_duration_seconds, and the range it used, so you can check the cut. The job costs $0.01.