Neural Amp Modeler (NAM) in HISE
-
@southfieldsound Not yet implemented but could be soon - follow this thread
Simple ML neural network
@orange Jatin said in the GitHub issue that he's thinking about adding the wavenet model to RTNeural, when that's the case, I'll resume the work to support N...
Forum (forum.hise.audio)
This is the latest development: https://github.com/jatinchowdhury18/RTNeural-NAM
Now we wait for Christoph to sound off on whether he can work with this.
-
@aaronventure that's great news. Just went through the whole thread -- I wonder if it would be possible to get this set up in HISE now, especially after the recent updates from Jatin. @Christoph-Hart would this be simple enough to add to HISE with Jatin's contributions? If not, would anyone be able to point me in the right direction as to next steps to get NAM distortion profiles up and running in HISE? Thank you!!
-
I've just started using NAM in my own music and it's seriously impressive
it looks like the RTneural implementation is using a pretty standard wavenet setup that works on a sample-by-sample level so it should be pretty easy to port over once it's ready, basically just a "more specific" neural node, it might even already be possible to use the regular neural node for it if you know how to import the weights
-
Any new with Neural Amp Modeler in Hise ? I have started training models on my excessive collection of guitar pedals, and i postulate - there is little to none difference. It is just awesome!
-
@Ben-Catman we're all holding our breath until January.
-
-
@aaronventure said in Neural Amp Modeler (NAM) in HISE:
@Ben-Catman we're all holding our breath until January.
What will happen in January?
-
@Fortune Chris will be resuming work on HISE.
-
@aaronventure actually Iβve already implemented NAM on a local branch of HISE but itβs not stable yet and quite CPU intensive. I hope we get the CPU down to competitive speeds with other implementations but that has indeed to wait until January.
-
@Christoph-Hart January 1st or 2nd?
-
In case it helps:
When I create Neural custom node and try to open it as Hardcoded FX, it doesn't work.
I had to embed the Neural network .json even with the Hardcoded FX too.
I thought maybe this was causing HISE to make extra efforts.
-
@Dan-Korneff more like January 17th..
-
-
Any updates?
-
@Sclass nope, working on other stuff.
-
@Christoph-Hart Still keeping an eye out for the NAM update :)
-
@Christoph-Hart Any news? This would be an amazing addition to HISE.
-
@Lurch I mean, it works? The performance is seemingly twice as bad as the NAM plugin, but only because the neural node in HISE is processing two channels, and the NAM plugin is processing one.
If you use the multi node to force mono processing and before that collapse the stereo signal into mono, you get roughly the same performance.
Create your NAM file in the colab https://colab.research.google.com/github/sdatkinson/NAMTrainerColab/blob/main/notebook.ipynb
Put in into the Samples directory.
const neuralNetwork = Engine.createNeuralNetwork("MidDrive"); const namModel = FileSystem.getFolder(FileSystem.Samples).getChildFile("model.nam").loadAsObject(); neuralNetwork.loadNAMModel(namModel);
Create a ScriptFX DSPNetwork and load the math.neural node.
It would be cool to be able to embed the .nam files into the plugin instead of having to install them separately.
-
@aaronventure said in Neural Amp Modeler (NAM) in HISE:
If you use the multi node to force mono processing and before that collapse the stereo signal into mono, you get roughly the same performance.
Haha wasn't all that drama about it being 8x slower than the NAM plugin?
It would be cool to be able to embed the .nam files into the plugin instead of having to install them separately.
Can't you just embed the JSON content of the NAM file into a script and it will be embedded in the plugin?
-
@Christoph-Hart said in Neural Amp Modeler (NAM) in HISE:
Haha wasn't all that drama about it being 8x slower than the NAM plugin?
I have no idea, I'm only now touching it for the first time.
I ran direct comparisons in Reaper and a Waveform Synth with NAM on stereo channels consumes 1% of CPU (20% of RT CPU), while two NAM plugins in total consume 0.9% of CPU and 18% of RT CPU.
So HISE is about 10% less efficient, which is good enough and entirely shippable if you ask me.
@Christoph-Hart said in Neural Amp Modeler (NAM) in HISE:
Can't you just embed the JSON content of the NAM file into a script and it will be embedded in the plugin?
I sure can. I feel silly now. Thank you.