Would appreciate some help from somebody. Opened a support ticket a week ago and got no reply.
I trialled SL Pro 7 a couple years back and got fantastic results using Unmix Stems. Resolved to buy at some point. Finally bought recently, now at v9, and before activating the serial thought I should trial version v9 to check compatibility. Was disappointed to find I didn’t get the same results using Unmix Stems on the same material. Evidently the algorithm has been tweaked, and for me, on this material, it’s much less effective.
I downloaded SL Pro 8 on a different machine, thinking maybe I could request a demo of that as well, to see if its results were in keeping with those of v7 or with those of v9. To my surprise it worked straight away, and I’m guessing this is because my trial license for v9 allows me to run earlier versions, but I’m not sure. In any case the results from v8 matched the results from v7.
I have the opportunity to buy a copy of SL Pro 7, which I guess may or may not upgrade to v8 on activation, but I have to confirm within the next couple of days if I want to buy it. My question is basically this: If I activate my copy of v9, can I run v8 perpetually using that license? If so, are there any particular steps I would need to take to get this working? If not, is there any way for a v9 owner to run v7 or v8, by adding a second seat or something? Or do I need to pick up this copy of v7 while I can, and try and sell my copy of v9? (In case it’s not obvious, I recognise that v9 is probably better, technically, overall, but I bought the software in order to do this particular work with this particular material.)
Of course, I’m also interested in knowing if there is something specific and straightforward that I’m missing with respect to the new Unmix behaviour – something that might explain the different results. But I don’t want to spend time fiddling with the settings. Part of what I liked in the first place was that the results were great right out of the box, although I did experiment a bit with sensitivity, finding that it didn’t make a material difference.
What I find with Spectralayers 8&9 unmixing audio layers is that the layers(especially the vocals and others layers) leans more towards the melody/harmony rather than percussion whereas the other applications (who have the ability to unmix) leans more towards the percussion. Maybe this has to do with the data sets and filters/gates applied, all-in-all every single application I’ve tested always fails (from the expensive applications to the free applications) to get a clean extraction. No application is perfect. I cant really say one is better than the other because each application uses different data sets with different algorithms and they all give different results.
@aztec Yes, with a SL9 Pro license activated you’ll be able to run SL8 Pro just fine.
However you won’t be able to run SL7 Pro with that license because SL7 Pro was using the old license system (eLicenser), while SL8 and SL9 use the new license system (Steinberg Activation Manager).
Yeah I’d like to know this too. FWIW I’m unmixing vocals and piano, specifically to get the vocal stems. Technically I suppose I’m removing piano bleed but there’s a lot of it. SL v7/8 gave me a very nice clean vocal stem, but v9 is tending to leave some piano material in there, and in some seemingly random places.
Interestingly, I already felt like SL erred on the side of attributing material to the vocal rather than to “everything else”. For getting the vocal stem I found SL v7/8 the best, and for getting a piano stem (removing vocal bleed from the piano track) I found RX 8 better, because the SL piano stem sounded hollowed out. I imagine this effect will be even more pronounced in v9.
One thing you could try : if unmixing a song at 48Khz, try resampling it at 44.1Khz first. Let me know if you get different results between SL7/8 and 9.
Thanks. The wavs I’m working with are actually all 16-44. I just tried experimenting with resampling before and after and I’m not seeing it make any difference.
I’ve trialled v9 on the same Win 10 machine I trialled v7 on. The attached images show the v7 output and the v9 output, on the same settings. You can kind of see the difference in the gaps.
