Summary

WAIVE uses public domain data, such as heritage archives, to train generative AI models and create innovative music tools for live artists.

WAIVE explores how public domain data can be used to create innovative AI-powered music tools. The project applies various AI technologies to transform audio and video recordings into ready-to-use music samples, MIDI patterns, and VJ content for live performances.

 

By working closely together with leading European Heritage institutions, WAIVE is able to process enormous amounts of public domain audio and video files, such as field recordings and public broadcasting. These are then used to develop a series of open-source DJ and VJ plugins (VST’s) that can be connected to common DAW’s such as Ableton and Logic Pro. In addition, live showcases and workshops are organized to share the experiences and knowledge gained from the project.

 

 

Challenge

 

In the world of generative AI, the use of stolen training data is a huge problem. It raises legal and ethical concerns regarding intellectual property and the exploitation of creative work by major tech companies. Our challenge is to explore if and how heritage material, as public domain data, can serve as a responsible and creative alternative. This approach not only promotes ethical data usage but also encourages the creative reuse of heritage materials – while developing user-friendly tools that challenge and inspire artists to push the boundaries of their creativity.

 

 

Result

WAIVE demonstrates that it’s possible to build valuable creative AI tools based on public domain data. It offers artists the opportunity to experiment and tell unique stories. In this process, the origin of the data is not hidden, but rather placed at the forefront, bringing cultural heritage to the stage. To showcase the potential of this technology and inspire others, WAIVE also results in a series of live performances where our VSTs come to life.

 

 

Way of working

WAIVE is a collaboration between Thunderboom, The Netherlands Institute for Sound and Vision, a diverse range of European heritage partners, tech companies and education institutes. The development process follows an agile methodology and integrates various perspectives and continuously refines its designs based on live testing and user feedback. This involves collecting and curating heritage data, building and refining out tooling, organizing live showcases to test our technology in real-world settings, and promoting community engagement through workshops and online knowledge-sharing initiatives.

 

Partners

Netherlands Institute for Sound and Vision, Arran Lyon, Superposition, BBC Broadcasting, Corporació Catalana de Mitjans Audiovisuals, Baltic Audiovisual Archival Council, King’s College London, Damn Loud Agency

This project is made possible with the help of Mondriaan Fonds and Stimuleringsfonds.