ABOUT THE TEAM

Jason Brancazio has been writing and releasing electronic music under various aliases since the mid 90s. He more recently learned to train and run machine learning models by building recommender systems for a well-known streaming media application. He entered the AI song contest specifically to intertwine his passions for music production and machine learning.

ABOUT THE SONG

Inspired by 90s big beat, my track is an exploration of co-creativity via sampling, source separation and style transfer with three pre-trained ML models: OpenAI’s Jukebox, Deezer’s Spleeter and Magenta’s DDSP timbre transfer model. I hope to demonstrate that generative models trained on existing recordings such as Jukebox can provide an infinite source of novel sonic material, spark endless creativity, and perhaps disrupt the sonic homogeneity of commercial EDM and hip-hop. I also hope to demonstrate a process by which an artist can responsibly, creatively, and respectfully sample audio from generative models.

ABOUT THE HUMAN-AI CO-CREATION PROCESS

Prior to the contest announcement, I was pursuing a simple question: Can AI generate funky breakbeats? After failing to train a competent VAE on a public drum audio dataset, I suspended my nascent research and instead experimented with the publicly released OpenAI Jukebox model. My goal was just to trick Jukebox into generating drum loops. To do so, I asked Jukebox to continue drum loops from a sample CD that I purchased long ago. I provided the model genres and artists likely associated with breaks and minimalist lyrics like “make it funky now!” that precede the break in some well-known funk songs. Jukebox produced great drum audio, but rarely for long, veering instead to deep and funky grooves. A natural idea emerged at the same time the contest was announced: I would use this material as I would have used samples back in the 90s to assemble a song. Jukebox samples are mono and gritty. Wishing to produce a higher-quality contest entry, I decided to use samples from Jukebox mostly as a scaffold for my song with the goal of removing as many as possible from the final mix. For example, I replaced a Jukebox-generated bass loop by transcribing it and re-rendering it with a higher-quality VST plugin. I also used the DDSP timbre transfer model to create a credible sax solo, and used Spleeter to separate vocals from loops. Source separation helped me add more character to the track and raise the level of the vocals in the mix.