Smart Controls & Song Structure

State A: Fresh Generation

When starting from scratch, you can guide the AI with:

Electronic Subgenres

Electronic covers a lot of ground, so Aether reads your prompt for a subgenre and writes to that style specifically: its tempo, its groove and its structure. Say what you want and you get it; say nothing in particular and you get house.

Naming a Genre in Your Prompt

You don't have to use the Genre menu. Mentioning a style in the message itself (“a slow jazz ballad”, “liquid drum and bass”, “a Chopin nocturne”) selects it just the same. The Genre menu wins when you have set it, so leave it unset if you want your words to decide.

Naming an artist works too. Aether recognises a wide list of artists as shorthand for the style they are known for, so “something like Four Tet” lands in downtempo electronica and “a Skrillex-style drop” lands in dubstep. It reads whole words only, so “popular” never triggers Pop and “greenhouse” never triggers House.

Your own language works too. “musique classique”, “klassische Musik” and the Japanese, Korean and Chinese names for a genre all reach the same place as the English one. The same is true of drum styles on the Drums tab. Not every style has a name in every language; where one is missing, the menu still gets you there.

And it answers in the language you wrote in. The reply, the composition plan and the rest of the chat come back in your language rather than English. What Aether writes is unaffected: the key, the scale and the drum piece names stay in their standard form, which is why the drum map in the example below still reads Kick, Snare and Closed Hat underneath a plan written in Korean.

Genre Sticks Across the Conversation

Once a conversation has settled on a genre, it stays there. You do not have to repeat yourself on every message, and a passing descriptive word will not knock the whole piece into another style halfway through a session.

To change it, name the new genre or subgenre directly: “make this a bossa nova instead” moves it. An artist name on its own does not: once a chat has a genre, “make the lead more like Adele” is taken as a note about the lead, not a switch to Pop. Until a chat has settled on a genre, an artist is enough to pick one, as above. Setting the Genre menu overrides everything, at any point. Starting a New Chat clears the memory along with the conversation.

State B: Editing Mode

When MIDI is already loaded, the panel switches to refinement controls:

Action and Restyle chips are multi-select: toggle several at once (for example, Reharmonize + Jazzy + Lush) to stack directives in a single pass. Aether flags musically contradictory combinations (such as Sparse + Lush) in real time so you can resolve them before generating.

Musical Parameters (Always Available)

Building a Structure

• ●Add sections in order: Verse, Chorus, Drop, Bridge, or anything you type. Names are passed to the AI verbatim, so non-pop terms like "Exposition" or "Recapitulation" work too.
• ●Give each section a bar count, or leave it at 0 to let the AI decide how to distribute length across the form.
• ●When Sections are active, the Bars field shows the running total of your structure so the arrangement length stays honest.

Genre-Aware Presets

Each genre supplies sensible structural starting points, so you don't have to build every arrangement from scratch: