ABOUT THE TEAM

Driven by collaborations involving scientists, experts in artificial intelligence, cognitive sciences, designers, and artists, the Metacreation Lab for Creative AI is at the forefront of the development of generative systems, whether these are embedded in interactive experiences or automating workflows integrated into cutting-edge creative software.

Team:

• Cale Plut (Composer and musician) is a PhD Student in the Metacreation lab, researching AI music applications in video games.

• Philippe Pasquier (Producer and supervisor) is an Associate Professor, and leads the Metacreation Lab. 

• Jeff Ens (AI programmer) is a PhD Candidate in the Metacreation lab, researching AI models for music generation.

• Renaud Tchemeube (Producer and interaction designer) is a PhD Student in the Metacreation Lab, researching interaction software design for creativity.

• Tara Jadidi (Research Assistant) is an undergraduate student at FUM, Iran, working with the Metacreation lab.

• Dimiter Zlatkov (Research Assistant) is an undergraduate student at UBC, working with the Metacreation lab.

ABOUT THE SONG

A song about the weekend (and you can do whatever you want) explores the relationships between AI, humans, labour, and creation in a lighthearted and fun song. It is co-created with the Multi-track Music Machine (MMM). 

Through the history of automation and industrialization, the relationship between the labour magnification power of automation and the recipients of the benefits of that magnification have been in contention. While increasing levels of automation are often accompanied by promises of future leisure increases, this rarely materializes for the workers whose labour is multiplied. By primarily using automated methods to create a “fun” song about leisure, we highlight both the promise of AI-human cooperation as well as the disparities in its real-world deployment. 

ABOUT THE HUMAN-AI CO-CREATION PROCESS

Throughout the creation process of A song about the weekend, we aimed to highlight the potential strengths of AI generation. Our main AI co-creator is the Multi-track music machine, or MMM. MMM generates new music, based on provided sheet music (in MIDI form). In addition to this AI, we use online generators for the song’s subject and lyrics. Finally, the singing and instruments are all synthesized by the computer. 

We first generated a random prompt: “A song about the weekend and you can do whatever you want”. We composed an example song based on this prompt, to be used as the input for MMM.

Composing this song allowed us to provide important musical elements and features such as “groove” or “feel” in musical form. 

Once we composed our input piece, we provided the MIDI to MMM and began our co-creative work. We used an iterative process with MMM - we provide basic parameters, MMM generates a new song, we listen, adjust our parameters and repeat. For the final song, MMM generated every measure for every part except for the guitar riff at the beginning and the background rhythm guitar chords. 

While MMM provided the notes and rhythms for our song, we provided some of the performance. We downloaded MMM’s generation as sheet music, and played the song back into the computer using a MIDI keyboard. This results in each note having a subtly different volume and release point, creating a human performance effect with synthetic instruments without altering the musical content.