Can I use Seed-Audio 1.0 output in a commercial game?

Output rendered through paid access routes carries commercial-use rights; check the terms of your access route. On Yif, the first paid tier includes a commercial license. Free-tier takes are watermarked and for evaluation only.

Use case

Game audio: the NPC and the room, one render.

A voice line without its room sounds like a voice actor in a booth. Seed-Audio 1.0 renders the bark and the place together: the innkeeper with the fireplace, the market crowd with the vendors, the storm with nothing but weather. For prototyping it replaces a pipeline; for shipping it drafts the whole soundscape.

Hear three game moments

The prompt pattern that works

A medieval tavern in a fantasy game. A gruff dwarven innkeeper greets the player: 'Back again, adventurer? The usual corner table is yours.' Fireplace crackle, a lute in the corner, mugs clinking.

Anchor the world first. One sentence of place ("a medieval tavern in a fantasy game") tunes voice, effects and reverb together.
Write barks as character + line. Species, temperament, then the quoted line. Rerun the same prompt for variations of the same bark.
Ambience-only prompts make loops. Drop the speech entirely for environment beds; the rain take above is one prompt with "no speech" in it.

Honest notes for game teams

Output is a mixed scene, not stems. If your engine needs separate voice and ambience buses, render them as separate takes (voice-only prompt, ambience-only prompt).
Takes cap at 2 minutes; loop or chain for longer beds.
No streaming API, so runtime dynamic dialogue is out of scope today. Batch-render your lines at build time.

Prototype a scene from your design doc

Paste a location description from your GDD into the playground and hear it. The prompt grammar guide covers cue sheet structure for multi-character scenes.