Questions, answered plainly.
Installing, formats, licensing, and how Signal Studio sits next to the tools you already use. Anything missing is a good forum thread.
Which build do I need?
Windows ships as a signed MSI, macOS as a notarized universal .dmg, Linux as a self-contained AppImage. All three are the same release; pick the one matching your machine on the Downloads page.
Do I need Python installed first?
No. Each build bundles its own Python runtime and the packages the processing cards depend on, so nothing is installed system-wide.
Is there a CUDA build?
The default installer bundles the CPU build of Torch. A separate CUDA variant is published alongside it for machines with an NVIDIA GPU.
Which file formats can it read?
The formats MNE reads: EDF/BDF, BrainVision, EEGLAB .set, FIF, and BIDS-organised datasets. See the data formats section of the docs for the full list and how each maps internally.
Does it send my data anywhere?
No. Processing runs locally on your machine. An account is only needed for the community forum and the instruments marketplace, never to open or analyse a recording.
Can I export results for further analysis?
Yes: recordings, epochs, and model outputs can be written back out for use in your own scripts.
Can I keep using MNE-Python or EEGLAB alongside it?
That's the intended use. Signal Studio is a visual layer over the same underlying methods, so pipelines built here stay legible to anyone working in MNE, and exported data goes straight back into your scripts.
How does it compare to commercial analyzers?
The methods are the published ones rather than proprietary reimplementations, the pipeline is visible end to end, and the licence carries no per-seat cost.
Can I extend it?
Yes. Cards are declarative and the extension layer is documented: see authoring a card in the docs.
Is it really free? What's the catch?
It's MIT licensed. Use it, modify it, ship it in your own work. The instruments marketplace sells optional skins; the application itself stays free.
Can I use it in published research?
Yes, including commercially. Citation details are in the repository.
Privacy
The application runs on your machine and does not send your recordings anywhere. There is no telemetry and no usage reporting.
It makes two kinds of network request, both visible in the source of the scripts that make them: it asks the release feed whether a newer version exists, and it fetches a card or a workflow when you install one from the marketplace. Nothing else leaves the machine.
On this website: signing in is handled by Clerk, which holds your account. Anything you publish to the marketplace or post on the forum is public, under the name on your account. There are no advertising or analytics trackers on the site.
Terms and licence
Signal Studio is published by NeuraCrypt and is free to download and use. The full end user licence agreement is shown during installation and is included with the installer.
The source repository is private for now, so the site does not claim an open-source licence it cannot point you at. If that changes it will be stated here and on the download page.
Anything published to the marketplace belongs to whoever published it. Publishing gives other users permission to download and run it; it does not transfer anything to NeuraCrypt. Third-party extensions are provided as is by their authors.
Signal Studio is research software. It is not a certified medical device and must not be used as the basis of a clinical decision.