rtneural plugin full demo

RTNeural Plugin is a SuperCollider/Max/pd ugen/external which uses the RTNeural neural engine for inference. This engine is designed for real-time safe inference of tensorflow and pytorch trained neural networks. It can be used to process real-time audio or to run inference at control rate. Networks can have any number of inputs and any number of outputs. Networks trained at a specific sample rate can run at any samplerate due to internal resampling using libsamplerate. 0:00 Intro 7:09 Downloading RTNeural_Plugin and Setting Up a Python Virtual Environment 15:06 Setting up Max/MSP and a MultiLayer Perceptron in Max 33:49 Setting up Pd and the MLP in Pd 43:47 Setting up SuperCollider and the MLP in SC 52:12 Training an MLP to add in SC 1:10:41 Loading Existing Models 1:20:47 Training an LSTM to be an Anti-Aliased Sawtooth Wave 1:37:21 Converting Models found on the Web 1:44:45 Training our own RNN distortion model The RTNeural_Plugin repository is here: https://github.com/spluta/RTNeural_Plugin Downloads of builds are on the Releases page. The link is on the right hand side of the repository. This project is based on the RTNeural inference engine: (https://github.com/jatinchowdhury18/RTNeural) I figured out most of this going through the GuitarML source code. Check out Keith Bloemer's tutorials on training networks here: (https://youtube.com/playlist?list=PLH-HGV0VF7tEKYKmFu_KPkk0fjrg7vVi2&feature=shared) inspired by the FluCoMa project: (https://www.flucoma.org) and Wekinator: (http://www.wekinator.org) See Ted Moore's excellent videos on machine learning in FluCoMa, here: (https://www.youtube.com/watch?v=t2AJumPX7o4&list=PLU5dOxELCKlYBVLPLwUdmCXeVdRrQzPch&pp=0gcJCV8EOCosWNin)