Setup

Amadeus requires an inference runtime and at least one compatible voice model before it can bake audio. The first launch opens the setup wizard.

Complete setup

1. Select an inference runtime reported as available.
2. Install or select at least one voice model.
3. Finish the wizard and create or open a project.

The available runtime choices depend on the computer and installed software. CPU execution is the most broadly compatible option. GPU and accelerator options require compatible drivers and hardware. If an option is unavailable, select CPU or correct the indicated system requirement before continuing.

Select a voice

A voice model supplies one cast for vocal rendering. Select a cast for each voice before baking.

Figure 1. The Casts dialog identifies installed voices and their available components.

Figure 2. The voice rail selects the active part and its cast.

Condition	Action	Result
A cast reports missing required components.	Install or repair that voice before selecting it.	The cast becomes available for baking when its required components are present.
The project contains several vocal parts.	Assign a cast to each voice in the voice rail.	Each part bakes with its selected singer model.
No voice is selected.	Select a voice in the rail.	The piano roll and Inspector change to that part.

Continue

• Editor indexes Plan, Timing, and Pitch.
• First Vocal gives the normal notation-to-stem procedure.