Introduction to Audio Synthesis with Machine Learning
This workshop will introduce Kapre, a framework for digital signal processing, and Keras, a high-level machine learning framework in Python. Participants will learn how to train a simple generative network, known a variational autoencoder, and use it to generate and export samples as wav files to be used in their favorite DAW or sampler.
Some Python knowledge is encouraged, though prior knowledge of machine learning is not required. As our development environment, we will use Google Colab, an online environment that offers free GPU’s for training our models.
Sample datasets for training will be provided, but participants may bring their own sample packs if they desire. Short, simple samples (percussive hits, synth leads) will yield the best results.
This workshop will be presented on twitch.tv/babycastles
*This program is supported, in part, by public funds from the New York City Department of Cultural Affairs in Partnership with the City Council.