← Back to Home Moises

Moises — the responsive session player

What Moises is, where it genuinely earns its place around Logic, and where the rest of this section takes you.

The concept in one line

Moises's role

SUNO invents the musical world from a prompt. Moises doesn't invent anything — it listens to audio you already have (a SUNO bounce, a Logic sketch, a recorded riff) and generates or isolates individual instrument parts that lock to that material's tempo, harmony, and groove. It behaves like a responsive session player reacting to what's already on the table, not a composer inventing a record from nothing. Logic remains the place where everything actually gets finished.

What Moises is good at

Separation

Finer than your DAW goes

Lead/background vocal splits, individual drum components (kick, snare, toms, hi-hat, cymbals), and guitar subcategories — breakdowns beyond what Logic's own Stem Splitter offers.

Generation

Parts that listen

The one capability nothing else in this chain has: generate a new part that reacts to your existing audio — a bass line that locks to the drums you already have, not one invented in a vacuum.

Round trip

A clean path into Logic

Stems come back into Logic by export/import — or straight through the Stems plugin on Pro — where comping, mixing, and mastering stay exactly where they belong.

Moises AI Studio's Stem Generation panel, mid-generation on an AI drum part.
AI Studio's Stem Generation panel — the generative layer, in the browser or desktop app.

The one thing to know before your first session

The export gotcha

Key, pitch, and speed changes silently do not carry into exported Separated Tracks — only the Audio Mix export bakes them in. There's no error and no warning, and you may not notice until nothing lines up harmonically in Logic. The Export Gotcha page covers the rule and the workaround — read it before your first real session.

In this section

The field guides

Eight guides, one method. Roughly in the order a reader meets them: how the field got here, how to work across models, then the tools themselves.