ACE Studio lets you direct an AI vocal, one phrase at a time

01From a recording to an editable singer

ACE Studio can turn recorded singing into MIDI notes and recognized lyrics, then synthesize a new vocal performance from that material. That makes it useful when a melody works but a word, note or phrase needs attention. This is a guide to its current workflow, rather than a new product launch.

If your starting point is a mixed song, ACE’s documented route begins with Stem Splitter: separate the vocal and instrumental, then convert the vocal to MIDI on a singer track. Select the appropriate language. Multilingual recordings need to be split into clips containing one language each.

02Clean up the words and melody first

Conversion gives you something to edit, not a finished transcription. ACE warns that recognized lyrics may differ from the recording. Its Vocal-to-MIDI documentation also cautions that instrumental bleed, delay and reverb can reduce accuracy. A clean vocal is the better starting point.

Check the words against the source and listen through the note transitions before adding effects. For a proposed short brand jingle, for example, this would let you correct a misheard product name while keeping the melody you wanted. Use a recording you own or have permission to adapt.

03Shape the delivery, then the detail

With V2 voices, Dynamic controls overall vocal intensity through a curve. Raise it for a stronger phrase or lower it for a quieter delivery. Breath & Phoneme lets you insert breaths, change their duration and adjust the timing of individual speech sounds.

Advanced controls expose Air, Falsetto, Tension and Energy. These are synthesis parameters, so moderate changes are the sensible starting point; ACE warns that extremes can produce unexpected results. The goal is an intentional performance, not maximizing every slider.

The interface depends on the model and project. V1 voices have separate documentation. Existing projects containing the older Power, Soft, Breathy or Chest edits retain those lanes, while new V2 projects use Dynamic.

04Build a small vocal ensemble

Unison combines multiple AI voices inside one vocal track. Offset changes their timing relationship, Spread adjusts stereo width, and Gain controls their levels. It offers a compact way to explore a doubled lead or a group vocal without manually assembling every singer on a separate track.

Older tutorials may call this Choir Mode: ACE says the name changed after version 2.1.1. Treat that as an interface difference when following a demonstration.

05Finish with familiar effects

ACE Studio supports third-party VST3 effects on Windows and macOS, plus AU effects on macOS. They sit in track or master FX chains alongside the built-in effects. Only 64-bit plug-ins are supported; on Apple Silicon, use native plug-ins.

That means a vocal can move from note and lyric editing to your preferred compressor or reverb inside the same project. Those plug-ins still need to be installed and available on your system. This workflow is most useful when you want to direct a specific performance, rather than keep generating whole-song alternatives.