Deep Learning in Audio for Absolute Beginners: From No Experience to a Deployed Model

Discover how to jumpstart your journey into deep learning for audio processing in this 40-minute conference talk from the Audio Developer Conference. Learn to debunk common myths, acquire essential skills, and create a virtual analog model of a distortion effect without prior experience or extensive datasets. Follow a step-by-step guide to set up a deep learning pipeline, train a neural network using PyTorch, and deploy your model in a real-time audio plugin. Gain insights into synthesizing datasets, understanding key concepts like loss functions and gradient descent, and applying your knowledge to practical audio applications. Explore resources for further learning and witness a live demonstration of the entire process, from setup to training a neural network for audio modeling.

Intro
Outline
Myths
Datasets
Virtual Analog Modeling
Neural Networks
Fir Filter
Recurrent Neural Network
Neural Network in Audio
Loss Function
Gradient Descent
Hyper Parameters
Epochs
Validation
Application
Spice Software
Code Base
Resources
Conclusion