What is Neural Audio Synthesis?

Registration

WORKSHOP IS FULL

Thank you for your interest!

Latent Soundings: A practical introduction to Neural Audio Synthesis

It’s not every day that a new audio synthesis technique emerges. With new machine learning we have the possibility of a neural synthesis!

In this introductory workshop we’ll be delving into the field of neural audio synthesis. We will trace its history interwoven with artificial neural networks and get our hands dirty with training machine learning models with your own sounds.

Audio synthesis with neural networks, also known as neural audio synthesis, is a research field within the larger field of AI that facilitates the development of generative sound synthesizers based on the features learned from an original audio corpus. Once ‘trained’, these neural network based synthesizers can generate new sounds which imitate the features and qualities of the original dataset as well as perform timbre transfer—a novel process that reshapes a new audio signal into the timbral qualities of the dataset learned by the algorithm.

We will explore techniques that allow us to combine two existing sounds in order to generate a totally unheard new audio. These processes open up a new way of conceptualizing aesthetic exploration within the process of computer music composition, sound design and sound art; shifting away from thinking of sound in terms of fixed media and towards thinking of sonic phenomena generatively.

To illustrate this, we’ll explore various sound art pieces that engage and explore conceptually with these algorithms and think about the dataset as a way to address broader socio-cultural topics.

The aim of the workshop is for participants to gain a basic understanding of the principles underlying these techniques and to incorporate them into their current workflow in sound design, music production, and composition.

We’ll be making use of existing open-source algorithms and delving into the philosophies behind them. As we do so, we’ll discuss the different types of implementations artists use in their work, as well as the ethical implications these technologies bring to the process of sound creation.

Participants should have a computer available for the workshop. No programming knowledge is required, but it’s always welcome.

Like all Intelligent Instruments Lab workshops, we are interested in a diverse group of workshop applicants (because diverse backgrounds are more interesting when generating new ideas), so if you are interested in the workshop but doubting whether you would fit into the workshop, you are the right person to join!

Workshop date and duration:

19 of June 2023

Time: 16:00-19:00

Location: LHÍ Laugarnes

Requirements:

Personal computer/Laptop

Facilitator:

Moisés Horta Valenzuela / 𝔥𝔢𝔵𝔬𝔯𝔠𝔦𝔰𝔪𝔬𝔰

As a self-taught sound artist and creative technologist from Tijuana, now based in Berlin, I navigate the realms of computer music, AI, and the politics of emerging digital technologies. My work, presented in spaces such as Ars Electronica and MUTEK, bridges ancient and modern sonic technologies and has earned me the A.I. Newcomer 2023 award. I lead workshops in creative AI, focusing on sound and image synthesis and the demystification of neural networks. Current projects include SEMILLA A.I., an interface for generative neural sound synthesizers, and OIR, an online platform for a semi-autonomous meta-DJ trained on extensive electronic music visuals.