In this workshop, we will cover the basic intuition behind the generative adversarial network and its applications in audio generation. We will use a variety of synth samples as training data and use the final trained generator to output a range of random wav files of unique sounds that can be used in a sampler or DAW.
In this workshop, students will learn:
- Basic concepts behind generative adversarial networks
- Introduction to WaveGAN
- How to build and train a generative model
- Use the trained model to generate and export unique sounds
- Prerequisite / Background knowledge
No prior background in machine learning is necessary. However it is recommended that participants have some familiarity coding in Python. All datasets, tools, and environments will be provided in the form of a Google Colab notebook, so only a web browser is required.
About the instructor: As a digital artist, Allan Pichardo finds the hidden things that modern technology obfuscates about ourselves and uses code to expose them in their naked form through speculative software—repurposed and hacked applications meant to subvert the original intention of the technology. Presented as interactive installations, netart, video games, and hacked software, Pichardo’s work intends to upend the relationship between consumers of technology and the power structures that exploit them for data.