I have been using Spectralayers to unmix old stereo tracks of my band playing live in a rehearsal space - pretty succesfully especially with SL10.
I listen to what is in the “Other” and “Non-Unmixed” layers and can hear portions of the drums, guitar and bass or whatever. Typically I combine these two layers and then “Unmix components”. The percussive layer goes to the Drums (before the “Drum Unmix” is used) and the “Tonal” layer is mostly split between the Guitar layer and Bass layer. I am left with the “Noise” layer which I don’t know what to do with. I feel for integrity I should either keep it in the mix in some way or perhaps add it to the drum layer before unmixing. Or simply delete it…
What do other people do? Does anyone have a better strategy or can make some useful suggestions? (I know - “it depends…” and “use your ears” but I am looking to learn from others useful and experienced comments rather than those if possible)
I tend to keep them. Sometimes, I feel a layer «miss» something and the «Other» or the «Non-Unmixed» seems to complete that layer. I see them as «useful artefacts» so to speak. Spectralayers as a hard time with evanescent sounds like guitar effects, electric piano sounds (Rhodes) and effects in general. Some of those effect or highly treated sounds ends up in the unmixed or other layer. It is not a simple task to make use of them I know.
You take a look at those layers and when in doubt listen.
In some occasions, obvious parts belong to another layer, which can be manually selected and sent to those.
Frequent synthesizer or other hard to classify sounds appear in these categories. As always depending in what you are doing with the files, is what you end up applying to those.
In some cases, a degree of useful sounds end there and there is always the possibility to go back and try another algorithm (or settings, remember that the Spectral part of Spectralayers (SL) actually process sound “visually” in function to what it is focused on: either on transients or in valleys)*1, so to find different ways to un-mix the original file, until you find the best SL process that suits your goals.
///Personally, I find those additional “other” and “Non-Unmixed” layers very useful, either to fine tune the un-mixing process, to later correct sound misplacements or to try other software with those (de-verb, de-filter, other spectral un-mixers, etc).
*Fortunately, when one changes a spectral setting (top right menu) the result is simultaneously displayed visually on the spectral time line, so one may appreciate how the setting being changed move its focus close or far from the sound we want differentiated or not.
Because I primarily record live gigs to mix and produce later on, the Unmix stem usually has some good data in it that should be kept. I don’t need to do any mixing with it, but it provides a bit of clarity and essential room reverb that would otherwise leave the recording sounding dull and boring.
The Unmix stem is where I find alot of saxophone is separated. There and vocals. Apparently the sax is the most instrument that most closely matches the human voice. So I will typically just rename it to Sax or Horns…
