>/ FAQ

ACENTAUR was not built to be a finished-song generator. While Text to Music is able to generate whole songs, the main purpose of this Instrument is to generate Track Candidates one at a time. You audition, reject, reshape or keep them, and every accepted track can become context for the next step in your Track Stack. You remain the producer.

ACENTAUR installs the original, unmodified model checkpoints of two families, and each keeps its own provenance and licence:

Stable Audio 3 (Small and Medium) is built by Stability AI, which states that its models are trained exclusively on fully licensed audio. Commercial use of the output is permitted under the Stability AI Community License.

ACE-Step 1.5 comes from the ACE-Step project, which states that its models were trained on licensed and freely available material as well as MIDI data. The datasets themselves have not been published. The model licence permits commercial use of the output.

EINAUX claims no ownership of anything you create with ACENTAUR. Your generations are yours to produce, release, sell and license. Whether purely AI-generated audio can be copyrighted varies by country and usually depends on your own creative contribution — selecting, editing, arranging and layering generations the way ACENTAUR is designed to be used is exactly the kind of contribution that strengthens your position.

The licence and notice files of every model family are installed with each model package, and the complete overview lives on our Transparency page.

Insert ACENTAUR into a plug-in slot on a track or bus. ACENTAUR receives the audio from that channel strip for capture and source-based workflows. You can also route audio to ACENTAUR through its built-in sidechain input. The available modes depend on the installed foundation models and appear only when their requirements are met.

Recommended for Logic Pro: To use Text to Music or Track Stack → Empty Stack, load ACENTAUR as an AU MIDI-controlled effect on a Software Instrument track. Logic can pause plug-in processing on Audio tracks without regions. If generated audio remains silent, use the Software Instrument setup described above. To capture existing audio in this setup, select the desired track or bus from Logic’s Side Chain menu.

ACENTAUR installs revision-pinned, release-tested local model packages rather than exposing a model repository inside the plug-in. The current provider stack includes ACE-Step 1.5 and Stable Audio 3 Small Music and Medium, running locally with MLX on Apple Silicon. Stable Audio provides Text to Music, Retake and Repaint; the broader ACE workflows remain available through the corresponding ACE packages. Local and open-model workflows are a permanent part of the product, while each model family retains its own licence and attribution terms.

ACENTAUR is a creative instrument, not a deterministic audio renderer. Each generation is a new interpretation of the source audio, prompt and selected settings. The same musical instruction can therefore produce different arrangements, performances and timbres.

Prompts guide ACE rather than issuing exact commands. Focused descriptions usually work better than long lists of competing ideas. Genre, instrumentation, playing style, mood and production character are useful directions, but none of them guarantees a specific result.

Occasionally, a generation may contain garbled passages, unexpected vocals, unstable structure, distortion or unusual spectral artefacts. This is a limitation of the underlying generative model. It does not alter your source audio, captures, history or previous generations. Discard the result, try another seed or batch variation, or simplify the prompt.

The Turbo models prioritise fast iteration. Generate several candidates and treat promising results as material to audition, edit and produce rather than as guaranteed final masters.

An 8-bar generation can sound like 6 or 7 bars. The file always has the exact length you asked for and starts on the downbeat, but the model composes an ending and often leaves the last one or two bars quiet. Ask for two bars more than you need and shorten the region in your DAW, or work in 4-bar phrases, which end cleanly far more often. This applies to Text to Music and to the first layer of an empty Track Stack.

For practical prompting strategies, Flow Edit guidance and troubleshooting, see Working with ACE: Prompts, Variations and Model Limits in the ACENTAUR Tutorial.

ACENTAUR comes as a native Audio Unit v2 and VST3 plug-in for Apple silicon Macs. It is tested and officially supported in Logic Pro 12, Ableton Live 12 and Universal Audio LUNA 2.

Core ACENTAUR workflows include audio capture, sidechain input, host-synchronised generation, project recall, local playback and file-based WAV drag-and-drop. Individual workflows can vary where a DAW does not expose Timeline clips or continuously process silent Audio Effect plug-ins.

Yes. We are currently working on AAX/Pro Tools compatibility on macOS. Keep an eye on the Change Log or email us if you want to be notified when the release is announced.

Universal Audio LUNA 2: Timeline clip drag-out
LUNA does not currently publish Timeline clips as files or macOS file promises during drag-and-drop. Timeline clips therefore cannot be dragged directly into ACENTAUR or other applications.
Workaround: Use ACENTAUR Capture for the fastest workflow, or export clips from LUNA with Shift-Command-K and load the resulting audio file into ACENTAUR. Generated WAV files can still be dragged directly into LUNA.

Ableton Live 12: MIDI Clip drag-in
Ableton Live does not expose instrumented MIDI clips to Audio Unit plug-ins as rendered audio through drag-and-drop.
Workaround: Convert the MIDI performance to an audio clip using Freeze and Flatten, Resampling, or Export Audio/Video, then drag the resulting audio clip into ACENTAUR.

ACENTAUR will load and generate audio in other hosts, that follow the official AU/VST3 implementation guidelines. Tested and supported are Ableton Live 12 and Universal Audio LUNA 2.0.

An internet connection is required for registration, the initial setup and model downloads. Generation is designed to run locally once the selected model package is installed.

Yes, but with one caveat: ACENTAUR Net renders on a second Mac, and every Mac carries its own activation. The trial activates only one; the full version activates two — which is precisely what ACENTAUR Net needs. If you have a second Mac and email address, you can pair a render server and a playback Mac using two trial licenses.

ACENTAUR requires an Apple Silicon Mac (M1 or newer), macOS 15 or later and 16 GB of unified memory. With 16 GB you can use Stable Audio, the ACE 2B package and, since 1.0.1, ACE XL Turbo, in every mode. The one thing 16 GB cannot do is Text to Music with the Medium or Large planner. Details, what each memory size gets you and the current limitations are on the Technical Information page.

The complete installation with all models, plugins, settings app and runtime needs 92 GB of storage. See Technical Information for more details.

Windows compatibility is part of our roadmap. But because of the siginificant hardware differences between Macs and PCs plus NVIDIA GPU this will take some time, so stay tuned. 

ACENTAUR 1.0 launches at an introductory price of €39 (regular price €69) for a perpetual license.