Generate instrumental music with the Suno API
Request instrumental music for a scene, a game environment, or an editing timeline. Start with the role the music should play.
Describe the scene and the arrangement#
A useful brief names the mood, instruments, energy, and context. For example: “Quiet exploration music for a woodland game, felt piano, soft strings, sparse percussion, unhurried.”
Request the instrumental explicitly#
Set make_instrumental to true. You can still use description mode; no lyric prompt is needed.
{
"custom_mode": false,
"mv": "sonic-v6",
"gpt_description_prompt": "Quiet woodland exploration, felt piano, soft strings, sparse percussion",
"make_instrumental": true
}Preview against the actual material#
Listen under the intended footage or gameplay. Prompting for a loop or a precise duration does not establish a guaranteed seamless loop or exact delivery length. Trim, mix, and test the result in your own pipeline.
Retrieve, compare and place the tracks#
Send the JSON above to the create endpoint using the authentication and UUID headers in Get started. Store the UUID with your scene ID and poll the task endpoint. When data[].audio_url is available, audition the returned tracks against the same scene before choosing one. A second generation is a separate paid request; polling the first task is not.
Adjust the brief for the scene#
For dialogue, reduce musical density and avoid a busy lead melody. For exploration, describe a restrained arrangement and stable energy. For a transition, describe the desired build. These are prompting directions, not guarantees of timing, looping, loudness or a particular result.
- Dialogue bed: soft electric piano, sparse bass, light texture, understated energy.
- Exploration: felt piano, warm strings, little percussion, unhurried movement.
- Scene transition: muted synth pulse, gradual lift, a restrained finish.
Use a repeatable scene audition#
Compare the returned tracks under the same scene at the same playback level. Save the selected track ID and a short decision note with the scene revision. A brief guides generation; final placement still requires listening and editing.
| Scene check | What to note | Next action |
|---|---|---|
| Dialogue clarity | Where music competes with spoken words. | Try a lower mix level first; revise the brief only if the arrangement still competes. |
| Energy and pacing | Whether the musical lift fits the intended edit. | Move or trim the track; request a new arrangement only when needed. |
| Loop boundary | Clicks, silence or an obvious musical jump at the edit. | Make and audition the edit in your audio pipeline. A loop prompt does not guarantee a seamless boundary. |
Store and deliver deliberately#
Keep each selected track’s task ID, title and URL with your project metadata. The service does not publish an audio retention guarantee in this preview. Verify that playback works before delivery, and confirm storage and usage rights before copying or distributing generated audio.