What is Suno Sample?
Suno Sample is an endpoint for making a new song out of one part of an existing track. You start with a clip from a previous Suno generation, send its ID to suno-ai/sample as clip_id, and mark the part you want with start_s and end_s in seconds. If you leave the window out, the first 60 seconds are used. The model then writes a fresh song around that sample, styled by your tags, with optional lyrics in prompt, a title, a make_instrumental flag, and negative_tags for styles to keep out. The model version is set with mv. The task returns a pair of new clips, and the source clip is not changed. Use it when one riff, hook, or chord loop in a draft is worth more than the rest of the song.
Key features of Suno Sample
Four features cover how sampling works on the API.
Choose the exact window
start_s and end_s set the sampled range in seconds and accept decimals. Line them up with a bar or a hook so the part you care about is what carries into the new song.
Sensible default window
Without a window, start_s is 0 and end_s is 60, so the first minute of the clip is sampled. Useful for quick experiments before you tune the range.
Restyle the sample
Set tags to move the sample into a new genre, add prompt lyrics, or set make_instrumental to true for a beat. negative_tags keeps unwanted styles out.
Two clips per request on v6
Pick chirp-hawk, chirp-hawk-wild, or chirp-goose with mv. Each task returns two new clips, and each one has its own ID for later edits.
Best for
Turning a hook into a full track
Sample the eight seconds that work and let the model write the rest of the song around them.
Genre flips
Take a section of a ballad and rebuild it as boom bap or house with new tags.
Beats around a signature riff
Sample a guitar or synth line and set make_instrumental to true to get a backing track.
Remixes with a recognisable motif
Keep one familiar part of a song so listeners can still tell where the remix came from.
Saving the best part of a draft
When only the chorus of a generation landed, sample it instead of throwing the clip away.
Use cases
A beat-making app can show a waveform of the user's Suno clip, let them drag a selection, and send the selection as start_s and end_s with a genre picker mapped to tags. A remix feature can sample the chorus of a released track and generate two genre flips for the artist to choose from. A batch pipeline can sample the first 30 seconds of every draft in a project and generate instrumental variants for stock music libraries. To trim a clip down without generating anything new, use the Suno Crop API, and for a looser, style-only reference try the Suno Inspiration API.
API examples
Sample shares one docs page with mashup and inspiration: Suno remix docs.
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "suno-ai/sample",
"input": {
"mv": "chirp-hawk",
"clip_id": "1901ec96-72c6-4c26-8e56-91028ca6070d",
"start_s": 0,
"end_s": 30,
"tags": "boom bap, vinyl"
}
}'
The call returns a task_id. Poll GET /v1/tasks/<task_id> until the two new clips come back, or set a callback_url next to input to be notified when the task finishes.
FAQs
People also ask
It takes one clip from an earlier Suno generation, cuts a window out of it with start_s and end_s, and writes a new song built around that sampled part. Each request returns a pair of new clips.
No. start_s defaults to 0 and end_s defaults to 60, so without them the first 60 seconds of the clip are sampled.
Set start_s and end_s to the seconds where the hook plays, for example 42 to 50. Both fields accept decimals, so you can line the window up with the beat.
Sample works from one clip and a specific time range, so the chosen part carries into the new song. Inspiration (suno-ai/inspiration) takes one to four whole clips as a general style guide.
Yes. Use tags for the new style (for example "boom bap, vinyl"), negative_tags for styles to avoid, prompt for new lyrics, and make_instrumental for a beat without vocals.
Yes, each request is billed because it generates new audio. The live rate is on the pricing page. Failed tasks are never charged.
