Steinberg has included a limited version of SpectraLayers in Cubase since version 13, and there is a 4.26GB “SpectraLayers 12.0.30” download available in the Cubase Pro 15 > Tools folder of Download Assistant. The limited SpectraLayers feature set includes “Unmix Song,” which is stem separation.
In Dom’s demo video, the new feature looks very similar to the SpectraLayers ARA Unmix Song interface, but it has fewer stem options and runs WAY faster than SpectraLayers does. This leads me to some questions:
- Does the new “Separate Stems” feature of Cubase use SpectraLayers, or is it truly built into Cubase now?
- Considering how fast it is compared to SpectraLayers, is the resulting quality as good, or does it suffer compared to SL? (Like, is it still worth it to use SL via ARA instead of the new feature if high quality is required?)
I know SpectraLayers does a lot more than just song unmixing. (I own the standalone product.) I am just really curious what makes the new feature different and whether it’s connected at all to SL.
As I understand, SL is the core code.
However…..
- Spectralayers one (or whatever it’s called) yanks vocals out of a mixed vocals/instrument mix
- The demixing routine in Cubase 15 lets you yank FOUR things (in general) out of a mix…..very vanilla spleeter-like stuff.
- Spectralayers pro lets you yank out 6-7 families of things AND split mixed drums into separate bd/sn/hh etc AND….lots of stuff beyond.
Pick your medicine. The choice is yours 
We, as users, don’t know, but the integration is complete, transparent and the process is no different than applying an offline process. The speed is way faster than SpectraLayers but you are limited to only vocals, bass and drums (and “Other”, which is what gets left over).
The quality is about the same as free and open source options, good enough for karaoke and basic rebalancing of instruments, and beat detection could be helped by having a separate drum track. It’s way better than e.g. RX 11.
Absolutely – SL is a whole other class, but slower and with a higher learning curve in order to achieve the highest quality results. If you already have SL and use it via ARA2 then the only advantage of the new Cubase 15 stem separation feature is convenience (speed and integration). Depending on the material, it’s probably good enough for many applications.