Keep AI Studio for custom renders, then move to the Store for catalog tracks, SFX, plugins, and licensing coverage on client work.
Grab a free starter kit — 50 sounds, no card.
Drum hits, one-shots, a few loops. Open in any DAW.Each persona is a curated synthetic voice with a default style and a city of origin. Pick one in the composer, or pin one as default in account settings.
We start from 8–12 minutes of consented studio takes — clean speech and a short sung phrase. WavLM features pull a 768-dim speaker embedding that the model treats as a stable anchor across renders.
The voice tag (alto, tenor, baritone) maps to a top-k sampling preset in the latent diffusion stack. EnCodec hands the waveform to Vocos; the persona keeps its tessitura even when the prompt drifts genre.
Each persona ships with a prompt prefix the composer never has to type — cadence, language bias, room tone. Pick one, write the verse, hit Generate.
