Hey, 44.1 khz Only Option makes this unusable for me. Any way around this?
I’m trying to run it in Cubase 13
Cheers
Wayne
It runs fine in Cubase on 44.1 kHz. Why do you think it’s not usable?
Sorry if that was unclear. I mean having my sessions be at 44.1 is not an option for me.
I have a similar problem with Live 11/12 because my audio device (an aggregated audio device) only operates at 48kHz.
My workaround was to run an A/B split on my guitar signal and feed two interfaces; A sends to my main setup and B sends to a 2nd audio interface which is set up under MG3 standalone at 44.1kHz/256buffer followed by a virtual audio cable. I use Loopback 2 (a paid app which 100% solves the issue) which is aggregated with my main audio device. It resamples the audio output at 48kHz and sends it through to Live.
You could probably achieve the same result with Blackhole or Virtual Audio Cable but I can’t verify if they will resample audio correctly as Loopback 2 does.
I’m not a Cubase user so please forgive my ignorance on how it works. Consider Logic Pro, which allows the project sample rate to be different to that of the device/s allocated in the audio settings. Does Cubase allow this independent sample rate setting or not?
1 Like
No, cubase would not allow different sample rates in one project.
If that is the case, then I believe that the OP would be best guided to try the method I outlined above, or use MG2 until a multi-sample rate/adjustable-buffer version of MG3 is released.
I’m aware this is a severe limitation for some studios and of course we will make any sample rate run as soon as we can. But I just wanted to add a note:
The background for this limitation is subtile: When samples come at different rates, we need to re-train our ML models for that. We cant just resample it because that will always add a bit of latency. It’s not a problem, and the end result will be very similar (but not completely identical) tracking. So, ideally, to manage my time effectively, we would add support for all samplerates when the tracking is frozen, right before the official launch of MG3. But I may give in for pressure and do it now, it’s just not optimal in terms of development time.
For the above reason I suggest to use this BETA on a hobbyist basis, on some laptop. I’s not yet the time to run a full studio on this.
5 Likes
Ok great to know. I’m a commercial/ Film composer. Everything is always at 48. If it’s possible to make that adjustment in the future that would be amazing!
Thanks for the reply and hard work
1 Like
Thanks for update. I am also awaiting higher sample rates as I normally work at 48kHz, but I am also very hesitant to ask you to alter the road map for beta testing. I think I would prefer to get a rock solid fast reacting monster 44.1 kHz version before retraining the models to other sample rates.
In the mean time a couple of related question, and feel free to pass on them if you feel the answers would reveal too much of the innards of MG:)
Is there likely going to be a gained performance in terms of tracking speed and possibly also accuracy with higher sample frequencies? I suspect this depends of if and by how much you are already upsampling the signal.
Were all the training data performed by the same guitarist or were more people/styles/guitars involved?
And, the million dollar question, could it possibly in the future - MG4 or 5 - be possible to train MG with your own datasets? The Neural Amp Modeler was a real game changer for me after it allowed me to sample my guitar amp and distortion with amazing results. I know that a training set for something like MG would be much more complex than 3 minute file of a recording of you gear with some weird bleeps and sweeps (like NAM) but but but…it would be so extremely cool and useful.
@WayneH - I hope it is ok I jumped your thread:)
It might be possible to fine-tune models with your guitar - that’s a good idea and something for later, maybe 
3 Likes
That would be amazing!
Thank you, I work at 48 khz as well. Cheers