Open Culture Tech + AIxDESIGN Making Immersive Tech Accessible

 

Open Culture Tech is an initiative by Thunderboom Records and the Netherlands Institute for Sound & Vision that makes immersive technology more accessible to artists through residencies, open-source tools, and educational showcases. As part of OCT 2.0, AIxDESIGN partnered to host a series of workshops exploring how AI shows up in music production—both its creative potential and challenges around agency, equity, and sustainability.

Meet Linalab

 

In October 2025, we hosted Barcelona-based musician, educator, and live coder Lina Bautista for a hands-on workshop exploring variational audio autoencoders at DOOR Open Space in Amsterdam.

 

 

Lina is a musician, artist, educator, and developer who combines modular synths, DIY electronics, and computers to make music and engage audiences in sound technologies. She's part of Toplap Barcelona and Axolot colectives, and teaches at universities in Barcelona. Her approach treats AI as a craft technique—designing tools for expression rather than replacement.

 

This philosophy shaped the entire workshop: rather than treating AI as a black box that generates finished tracks from text prompts, Lina guided participants through the inner workings of neural audio synthesis, showing how to manipulate sound at a fundamental level.

 

 

Workshop announcement

• AIxDESIGN on Instagram: "WORKSHOP: Transiting the Latent Space w/ @linalab"
• Event RSVP: https://www.dooropenspace.com/program/gwm-rf6tz

 

 

 

A Brief History of AI Music Generation

 

Lina began by contextualizing RAVE within the history of AI music generation, walking participants through different deep learning architectures—from the sequential processing of recurrent neural networks to the attention mechanisms of transformers and the noise-reduction approach of diffusion models. This grounding helped everyone understand not just what RAVE does, but where it fits in the evolving landscape of AI-assisted music creation.

 

Recurrent Neural Networks (RNNs): Designed for processing sequential data where order matters—like melodies unfolding over time. DeepBach (2016) used this approach to generate convincing four-part chorales in the style of J.S. Bach.

 

Transformers: The "attention is all you need" architecture that powers much of modern AI. In music, this means tools like MusicGen (Meta) and MusicLM (Google)—the text-to-music generators you've probably encountered.

 

Diffusion Models: Starting from noise and gradually refining it into structured audio. Examples include Harmonai's Dance Diffusion and Riffusion, which applies image diffusion techniques to spectrograms.

 

Variational Autoencoders (VAEs): This is where RAVE lives. Instead of generating from text prompts or refining noise, VAEs learn to compress audio into a compact "latent space" of manipulable variables, then reconstruct it back into sound.

 

 

 

 

What Are Variational Audio Autoencoders?

 

 

At the heart of the workshop was RAVE (Realtime Audio Variational autoEncoder), an open-source tool developed at IRCAM.

 

RAVE encodes audio into latent space—think of it as a compressed mathematical representation of sound's essential features. You can then navigate this space, morphing between different sonic characteristics, discovering sounds that don't exist in the training data. It's not about convenience or automation; it's about creative exploration.

 

This latent space is where the magic happens—it's like a secret map of sound's hidden dimensions. By navigating these dimensions, you can go beyond simple timbre transfer or voice cloning to discover entirely new sonic possibilities.

 

This approach resonated with participants who were tired of cookie-cutter AI music generation. As one participant put it, this was a "completely different approach to AI music" from the text-to-music tools flooding the market.

 

 

The Hands-on: Navigating Latent Space with MAX-MSP + nn~

 

The workshop used MAX-MSP visual programming and the nn~ external object to work with RAVE models in real-time. Participants explored pre-trained open-source models from both IRCAM's official repository and the Intelligent Instruments Lab (III) collection on Hugging Face.

 

Participants learned to encode audio into latent space, manipulate those variables in real-time, and decode the results back into sound.

 

Throughout the session, participants experimented with different models trained on various datasets: percussion, classical instruments (Musicnet), voices (Isis database), vintage music, and more - with each model opening up different sonic territories to explore.

 

 

 

Three Core Patches:

 

1. Sound-input-percussion: Participants fed different samples through a RAVE percussion model, comparing the original input with the reconstructed output. This helped everyone hear what the encode-decode process does to sound.

 

2. Sound-input-voices: Working with a vocal model that had many more latent variables than the percussion model, Lina encouraged experimentation with removing some of the connections between encoder and decoder, and replacing them with other signals to see how it affected the output.

 

3. No-input-guitar: The most experimental patch—no audio input at all, just directly manipulating the latent space variables. This is where the "transiting" happens: navigating through the space itself to generate entirely new sounds.

 

 

 

Links & Resources

 

Primary Platform:

• MAX-MSP for visual programming
• nn~ external object for neural network integration

 

AI Models:

• Official RAVE GitHub Repository
• RAVE models from IRCAM (Percussion, Musicnet, Isis voices, Vintage, Sol, VCTK)
• III community models on Hugging Face

 

Materials:

• Sample libraries from Freesound and Tidalcycles Dirt-Samples
• Workshop Materials, including patches and models, are available on Lina's website.
  • Workshop slides

 

Inspiration:

• Emptyset's "Blossoms" album as artistic reference as it was created using similar techniques
• Canblaster performance for IRCAM Forum Studio Session using RAVE

 

 

Reflections

 

Alternative approaches to AI Musicmaking

The group was diverse—a musician, a handful of producers, a creative coder and educator, a live music coder, and a new media art student—bringing different perspectives and experience to the experiments.

 

The hands-on nature meant everyone left with practical experience: understanding how to install and use RAVE in MAX-MSP, or Pure Data. More importantly, they left with a new way of thinking about AI in their creative practice.

 

Participants shared how the experience opened up new ideas on how to interface with sound and described the workflow as "very unique"—unlike and even counter to anything they'd encountered in other AI music tools.

 

 

ADE vibes + Clubcult exhibition

 

The workshop took place during Amsterdam Dance Event (ADE) at Door Open Space, which meant other exhibitions and performances happening in parallel to the workshop.

 

After the workshop, participants could check out the beautiful Clubcult exhibition—a beautiful portrayal of the final weekend of De School (the iconic Amsterdam club) and a hypnotic ode to rave and clubbing culture generally.

 

The other exhibition was about the graphic design flyers and promo materials of Dutch rave culture—Max from OCT/Thunderboom was also involved. It was a fun time travel through parties we recognized and were ahead of our time. Delightfully chaotic, it added beautifully to the atmosphere and ADE vibes.

 

 

 

What's Next

 

This was the second workshop in our OCT x AIxDESIGN series exploring AI for musicians. Thank you Lina for bringing such depth and clarity to these tools, and for championing open-source, ethical approaches to AI music technology!

We're continuing to host both in-person and online sessions for music makers and sound-adjacent creative communities. Keep an eye on our channels for upcoming workshops.

The "Transiting the Latent Space" workshop was hosted by AIxDESIGN and Thunderboom Records as part of Open Culture Tech 2.0, supported by Cultuurloket DigitALL in collaboration with Sound & Vision, Superposition, Wij Doen Dingen, Bureau Moeilijke Dingen, Bedrijf de Liefde, Chagall, and Wesley Hartogs.

 

 

Learn More →

 

• Open Culture Tech
• AIxDESIGN
• Linalab