I use Pro Tools for audio editing, and because of the nature of my work, I often process entire audio files that are over an hour long without splitting them into smaller clips first.
The problem is that processing often takes 30 minutes to well over an hour. The modules I use most frequently are Unmix Song (High) and Unmix Noisy Speech.
Lately I’ve been wondering if upgrading to an M5 Max (18-core CPU / 40-core GPU) would significantly reduce those processing times. Would the performance improvement be large enough to justify buying a new MacBook?
Also, does SpectraLayers still not take full advantage of the Apple Silicon GPU, or has that improved?
I’d really appreciate any advice or real-world experiences from people who have upgraded. Thanks in advance!
If you want to Unmix in SpectraLayers and don’t want to use other tools like Logic or Ultimate Vocal Recover, you basically need to get a Windows PC with a 24 GB (or more) VRAM Nvidia GPU.
this processing time on a Mac is a pretty big drag and workflow killer. It doesn’t seem any faster as a standalone process then as ProTools ARA. How is this possible that on a modern M series Mac processor that this is so slow?
I just tried using the stem splitter in logic. It’s so much faster and sounds great. It actually took me less time by far to drag the ProTools files into logic, process them with the stem splitter, export tracks as files, and import and spot the processed Audio than it did to use ProTools ARA and spectral layers. I mean far less time. what took me 12 minutes to process in Spectral Layers ARA took me less than five to go through this whole export to Logic and process loop. I sure hope Steinberg gets their act together on this one.
Apple has unrestricted access to Mac hardware, whereas the developers of SpectraLayers do not. Robin wrote in another thread some time ago that he had asked Apple every year to remedy this, but they essentially ignore him.
What would help are two separate versions of SpectraLayers (one for Mac, one for PC), but that would basically double the development effort, or the usage of Python, which Robin doesn’t want to use.
I understand what you’re saying. But I’m pretty sure Apple offers a tool kit to help developers integrate their software to work on the Mac platform. Plenty of other vendors don’t seem to have a problem having their programs run efficiently on the new M1 processors. This sounds like a decision that this Robin person has made to not put in the time. That’s their prerogative. In the meantime, I’ll use another companies program to solve the problem.
Yes, I know. I’ve only seen two YouTube video reviews, and they didn’t show any improvements over the visual GUI.
So – I don’t know if the models have been improved, and I don’t want to be a guinea pig when the price of the upgrade is so high.
ReStem 2 is fantastic! So much faster than ReStem1, but you need at least 10GB of video memory. 8GB is not enough - with an 8GB card, highest quality separations take around 8 mins for a 4min song. The GPU is partly used, but not fully, as one of the processes requires more than 8GB, so it’s slow, but still faster than ReStem 1.
With a 24GB card (which I have) separations take 1 minute. I’m not sure what the minimum requirement is currently for 100% GPU renders.
Having ReStem available within ARA workflows is really useful!
(The 8GB “limitation” is being worked on by Rim as it was not expected, but no date for release of a fix)
Coming to this app months after release, I have to say I was expecting more from an app that costs more than others. As a Mac user on an M4 Mac with 128 GB of RAM, I was expecting more. It’s slow and clunky, and the lack of module tweaks is very disappointing, and I find other Audio Splitting apps do this better and faster. I just think there should be a way to leverage more of the power of the mac inside this app (While in standalone mode).
