DocsAPI reference

OpenAPI specification

Model IDs and options in examples are illustrative. Select an available model from GET /v1/models and check its capabilities and current prices.

Music

Music uses the same job flow at /v1/audio/music. Create a job with a prompt, poll it, then download the track and, when the model writes them, its lyrics.

curl https://ark.dewaweb.com/v1/audio/music \
  -H "Authorization: Bearer ark_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: jingle-001" \
  -d '{"model":"google/lyria-3-pro-preview","prompt":"An upbeat acoustic pop song about a rainy afternoon in Bandung, about two minutes"}'

curl https://ark.dewaweb.com/v1/audio/music/MUSIC_ID -H "Authorization: Bearer ark_live_YOUR_KEY"
curl https://ark.dewaweb.com/v1/audio/music/MUSIC_ID/content -H "Authorization: Bearer ark_live_YOUR_KEY" --output track.mp3
curl "https://ark.dewaweb.com/v1/audio/music/MUSIC_ID/content?variant=lyrics" -H "Authorization: Bearer ark_live_YOUR_KEY"
ParameterTypeDefaultDescription
modelstringrequiredAn available music model.
promptstringrequiredGenre, instruments, mood and length, or your own lyrics with [Verse] and [Chorus] tags. Up to music.max_prompt_chars.
response_formatstringmp3From the model’s music.formats.

Each song is charged per track at the model’s price, held when the job is created and settled when the track is stored. Prompts that ask for a specific artist’s voice or copyrighted lyrics are refused without a charge. Tracks are kept for 7 days. A song is written in one call, so DELETE /v1/audio/music/{id} cancels it only while it is still queued; once generation starts it answers 409 job_not_cancellable.