SpectraLayers 10.0.30 Maintenance Update

Dear all,

we are happy to announce the immediate availability of a new patch for SpectraLayers 10, version 10.0.30.

Besides a few issue resolutions throughout the application, this update introduces traditional and and simplified Mandarin support for the GUI and other improvements.

For a detailed list of all improvements and issue resolutions, please refer to the version history.

The installers are available to download from the Steinberg website and via the Steinberg Download Assistant .

All the best,


But no Apple Silicon GPU support yet. Sigh.

The plan was to implement that in this patch…
Sad it didn’t happen. :disappointed:

@nikope @Marshall The improvements to the ONNX library I was waiting for came late (just a week ago). This could allow some form of acceleration for some specific AIs (but not all, due to the same CoreML restrictions).


Was hoping EQ match preset storing the spectral profile as stated in a previous thread ? Hope it will come for the next one.

@Stern unfortunately not enough time for this patch, it has been rescheduled

Thanks Robin for letting us know. Just to get a perspective on the CoreML restrictions: are these restrictions temporary? Or has Apple set them “in stone”? And if so, would this imply that users with Apple Silicon will never enjoy the accelerations that Windows/Intel/AMD users already have?

Thank you for the feedback! I appreciate that there are more frequent updates with V10 now.


Thanks for the impressively quick turnaround on these updates. However, is there any real reason to update to this latest version if your gpu is already supported?



@Magnus_N It’s a pretty minor update, and the other fixes/improvements are related to Process > Imprint, transcription/markers CSV export/import and Edit > Transform. If that’s not features that you use, you may skip the update. In theory there might also be a slight performance boost for Voice Denoiser and Unmix Noisy Speech, but that’s probably negligeable.

1 Like

@nikope Unfortunately for now it’s rather “in stone”. As it is, CoreML doesn’t allow data blocks longer than 16374 samples, which is a blocker for a lots of audio AIs.
It doesn’t mean that there won’t either be GPU acceleration on mac (and actually a few are already GPU accelerated on mac, such as Unmix Drums and Unmix Components), but it makes that path harder than on Windows. Some AIs can work around that limitation (not all requires large blocks, such as the 2 mentioned above), but ideally it’d be best to have Apple remove that limitation.

1 Like

Thanks for the quick reply and confirming my hunch!



This is really unfortunate but thanks for the info, Robin.

1 Like