Hello!
It's so exciting to discover this library, thank everyone for the effort!
It might be a dummy question, but - could you please tell how load models? SmartPedal has the dropdown interface option, what about amp?
Thank you!

When using Reaper in Windows 10 (unconfirmed with other DAWs) the GUI locks and won’t switch to other plugins in the FX view. Observed that this is not a problem in Reaper on Linux Ubuntu.

I've started using SmartGuitarAmp and SmartGuitarPedal in some music projects and one of the biggest features I want is to have the plugin state be retained across DAW sessions, so that I don't have to keep reloading models and adjusting parameters when reopening a saved project -- I can imagine that other users could find this feature useful as well.

Happy to tackle this myself! The main question I have is whether it'd be better to save the model JSON directly as part of the state (which would remove reliance on external files but increase file size) or just the model path (which would save space but break if the model file gets moved/deleted). I'm leaning towards the latter, since for now it's simple enough for users to reload a new model if they happen to move or delete them after saving. Curious what folks feel like would be best though, or any other solutions I'm not thinking of.

It'd probably be good to add similar functionality to the SmartGuitarPedal as well, but figured we could keep discussion to a single issue for convenience.

Currently, the knobs reset to the default after closing and reopening the GUI. Add state ability to save the knob states.

Hey Keith,
thanks for the VST!

i'm wondering how the parameters are implemented under the hood

with profiling (looking at you, Kemper) you have a snapshot of an AMP with set settings like Gain, Treble, Bass, Middle but the profiler itself can not accurately simulate the dynamics AMP at a higher or lower level of those settings - instead it applies post FX and EQ

with modelling you have a model of an AMP and when you tweak the settings it's more like an invisible hand is turning a knob on a "real" amp

which approach is used and do you have any ideas to improve the quality?

The little I have used this plugin I am super impressed with the sound, responsiveness when paying, etc. I do have some questions though if I may ask:

1. Do every model have one clean and one driven section? If not will the one it is does not have be silent if chosen?

2. Are the models offered using cab emulation or do we need to add that with for example another plugin?

Thanks in advance!

Hi there, SGA sounds cool !

Since we've added it to our collection for the HyMPS project, we strongly suggest you to check out other in order to establish some kind of collaboration that certainly could benefit all:

• https://sourceforge.net/projects/scorchcrafter/
• https://github.com/resonantdsp/SwankyAmp
• https://github.com/apohl79/GuitarAmp
• https://github.com/kaktus3000/HighGain
• https://github.com/andrepxx/go-dsp-guitar
• https://github.com/forart/ayemux

Hope that inspires !

The default GUI is large, would be nice to be able to resize.

I installed and tried out both the application and the vst3 plugin (1.0+git14.efe4597 ) for OpenSUSE Tumbleweed but could not find the button to load models like in the Windows version. Is there something I have missed or am do wrong?

OS: Debian Bullseye
JUCE Version: v6.1.1

I got this error when trying to build SmartGuitarAmp

Compiling PluginEditor.cpp
../../Source/PluginEditor.cpp: In member function ‘void WaveNetVaAudioProcessorEditor::loadButtonClicked()’:
../../Source/PluginEditor.cpp:271:17: error: ‘class juce::FileChooser’ has no member named ‘browseForFileToOpen’
  271 |     if (chooser.browseForFileToOpen())
      |                 ^~~~~~~~~~~~~~~~~~~
make: *** [Makefile:250: build/intermediate/Release/PluginEditor_94d4fb09.o] Error 1

it builds successfully using JUCE v6.0.8

Is this possible? I'm not a programmer by trade, the last time I messed with Python it was an hours long debacle. I think you'd see a big leap in attention if people could experiment with trying to train their own models...

It's an open-source plugin manager called StudioRack:
https://github.com/studiorack

The aim of the project is to provide a free open-source way to manage plugin dependencies and their versions. It would involve the following steps:

Publish a Github release containing the plugin source along with metadata plugins.json
Tag your Github repository with studiorack-plugin so it can be discovered
Wait for the studiorack-registry to index your plugin (every 24 hours)
Check the registry feed for your plugin to appear

automatable parameters are the only way for a blind user to do anything. Currently, the blind can only load the plug and process the signal on the default parameters. Nothing more.

Thanks for making this, under an open license!
I'd love to use it, but am using linux.
Since it's based on JUCE it should not be too hard to port it.
Any chance on that?

what's missing is a format for packaging amps
currently, you have to rebuild the binary if you want your own model + overlay

we could just zip our models and skins under a flashy extension (.amp?) and browse them in the plugin
appreciate your time and feedback

What are the chances on this running on Android and iOS?

If you need it, please contact me.
mail: clayidols#foxmail.com

Installed the CUDA-less Pytorch,

... I'm getting

File "C:\Users\Chip\anaconda3\lib\site-packages\pytorch_lightning\trainer\distrib_parts.py", line 317, in sanitize_gpu_ids
raise MisconfigurationException(f"""
pytorch_lightning.utilities.exceptions.MisconfigurationException:
You requested GPUs: [0]
But your machine only has: []

I thought I would try it with the "no CUDA" pytorch lightning but I guess I was too optimistic?

Any mac vst version in the making?

-move image loading outside of paint()
-only reload clipped parts of image

Apologies if redudant.

Downloaded the vst recently, put it into Ableton and it's using 80%-105% of my CPU. I have an older i5, but I've never encountered this problem with any VST before. Using the same methods too, placing it in the requisite folder, loading it in, placing it in the effects chain. When I turn it off, the CPU usage obviously drops.

Not sure what's causing this.

is there anything in the domain of machine learning for controlling a synth with a guitar, making a guitar sound like a synth, or turning the guitar signal into midi in a live setting?

Hi, I created this to install this plugin on Arch Linux (it's my first PKGBUILD!) and it looked like it was successful. I tried loading the plugin in Ardour and the GUI loaded, but the output was garbled. I'm not sure how to diagnose this and was hoping I could get some help with fixing this issue.

Currently, the WaveNet inference implementation has 5-6 times more CPU usage than a comparable commercial plugin. Investigate ways to increase the WaveNet algorithm speed while maintaining sound quality.

I couldn't get Reaper 6.67 Win10 to see both until I enabled this setting: