Audio detach limits: 1800 s source, 900 s output, wav or mp3

Audio detach extracts the audio track of one Sume-hosted video as a new wav (default) or mp3 for $0.01 a job. The source video may run up to 1800 seconds and the output up to 900 seconds, so a track longer than 900 seconds needs a range. The video itself is untouched. Details are in Audio detach.

The options

Which format

Wav is the format other Sume steps read: the timeline audio input, the Timeline audio route and speech to text all use it. Choose mp3 only when you are handing the file to a person or a player and file size matters. For transcription, pair 16000 with mono; the docs call that the STT shape.

A request and its guardrails

The source must already be on media.sume.com in your workspace. The server will not fetch from the open internet, so import first with POST /v1/media-imports. An Idempotency-Key is required.

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-mp3-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4",
    "format": "mp3",
    "range": { "start": 60, "end": 360 }
  }'

Cost of splitting a long file

A 25-minute recording needs two detach jobs of at most 900 seconds each, so $0.02. If you want many ranges from one track, detach once and split with Timeline audio instead of paying per range.

Sending ffmpeg-style fields such as filter or codec is rejected with ffmpeg_fields_rejected; the server builds the command itself.

Refusals to expect

Before detaching, probe the clip with video inspect and check probe.has_audio (frames: false is enough); a silent source fails rather than returning an empty file. Plan the range from the probed duration so the request never trips the codes below.