AI death metal band, Dadabots—developed by artists and mad scientists CJ Carr and Zack Zukowski—finds yet another way to push the furthest edges of creativity with their new epic, “RUG THE BOTS.” Combining what they call “human learning” with AI-generated choral harmonies, MIDI, and drum breaks, “RUG THE BOTS” stands in a category all its own.
Inspired by cornerstone articles of Ethereum lore, “Ethereum is a Dark Forest” and “Sandwich Attacks vs Salmonella,” "RUG THE BOTS” tells the story of taking revenge on malicious bots acting as ecosystem predators. “I wanted to make it something Ethereum themed,” says CJ. ‘“Ethereum is a Dark Forest” is the idea that hackers are lurking, waiting for users to mess up, and take advantage of them. Predator bots lurk in mempools, preying on you for profit, eyeing vulnerabilities and frontrunning opportunities, waiting to attack. But bots have weaknesses too. “Rugging the bots” is when you get sweet revenge and scam the scammer. The monster that terrorized your village? Now you're cooking it for dinner.”
Presented in the style of Old Testament religious zeal, the lyrics and vocal delivery evoke a crusade battle cry.
My Lord, O my strength!
I bless You for the faith You’ve had in me
My God, My King,
With Your grace,
First You give me weakness
And then lead me to prosperity
My faith, MY faith in You is never broken;
And he who knows not his ungodly ways shall be stricken down in Your wrath.
King of kings
Lord of lords
The Sword of swords
Cast from me my enemies into the pits of fire
“To make the lyrics, I used a technique I call ‘human learning.’ I had my roommate Erik Leone, who's a great musician and thespian, read the Psalms of David for a week, and then he improvised on the mic for five minutes. It came out way better than any AI model,” CJ laughs.
A parody twist on the concept of AI training–using the apocalyptic sermons of the Psalms of David as the training data, and Erik as the generative model–“RUGS THE BOTS” is a true novelty. “It’s funny because that's the typical process for making AI music: having training data that the machine is training on. The machine learns to imitate its style and then can improvise new material even infinitely, like on the death metal live stream. It’s been running for four years,” says CJ referring to Dadabots’ 24-hour AI-generated death metal livestream on Youtube.
Staying true to Dadabots form, “RUG THE BOTS” also incorporates various AI-generated elements including dog barking sounds and remixes of symphonic harmonies generated from previous Dadabots project “Bach Faucets”–a collection of four-part choral harmonies trained on Bach, minted as part of BrainDrop’s "All AI Art Looks the Same.”
“From the early days of David Cope's Bach By Design to DeepBach, BachBot, BACHPropagation, Bach 2.0, the Bach Doodle, Google Coconet, OpenAI MuseNet, Aiva, Amazon DeepComposer, and so on, Bach has been re-emerging out of AI, poured high from the chutes of the Baroque era, down the steps of the ivory tower, spurting through Big Tech research demos, and flooding our ears,” says CJ. “This phenomenon is so rampant a term, ‘Bach Faucets’ was coined in its honor by Dr. Kate Compton. I love the term and kind of ran with it.”
Fusing this Baroque-era classical style with drum’n’bass breakcore, “RUG THE BOTS” is of the Baroquecore genre–a niche musical style created by artist Igorrr. Dadabots has experimented with Baroquecore previously in their collaboration with Holly+, I Throw My Things Down the River and their 2022 submission to an AI song contest, Nuns in a Moshpit.
The percussion in “RUG THE BOTS” was made using a new tool Dadabots is currently developing that generates drum loops, amen breaks, bass synths, and foley from text descriptions. “The drums are an output from the AI model. And then to make it even more extreme, I brought up the tempo and started cutting the loops up into lots and lots of tiny slices. It’s hyper-frenetic.”
“RUG THE BOTS'' also employs Coconet by Anna Huang to generate MIDI, a tool for composing baroque harmonies in MIDI by generating accompaniment from an input melody. The song also features a choral section sung by CJ, accompanied by the otherwise AI-generated choir. “What I love about generating assets using neural nets is that I get so many more colors to paint with. I don't have to stick with one genre, I can make all these fusion genres.” Haunting and manic, the song’s nearly six minute journey evokes sitting in a church pew from hell. Overstimulation at its most theatrical.
Ten years into the project, Dadabots still manages to find ways to innovate on their own inventions. Their story begins at Berklee College of Music in 2012 where CJ and Zack met while interning and bonded over their mutual love of extreme genres of music. “Contrary to popular belief, we’re both musicians,” says CJ who used to be the lead singer of a metal band and a mathcore guitarist.
They also shared a love for computer science and its creative capabilities, CJ being a prodigy coder who started programming games at age 12 (his first games on Newgrounds have been played by over 2 million people). “I've always been creative and liked drawing and building with cardboard, but computers had the most promise for unlimited creativity. It was a no-brainer to try to figure out how to use them.”
Building a formidable creative partnership with Zack, who is a researcher, hacker, and multi-disciplinary artist, Dadabots was born. “I've always been into the fringes of extreme music as far away from typical 4/4 radio friendly music as possible, pushing music as far as I can with instruments and electronic production. I realized I was at a limit of how far I could push that and was going to have to use algorithms,” says CJ.
CJ and Zack began Dadabots by experimenting together on their first project–a SoundCloud algorithm remix bot. Dadabots took on a new level of intention in 2015 when deep learning took flight. “I quit my job programming advertisements. I'm like, ‘Finally, we're here at this moment where people are using neural nets to make art.’ My professors in school had always been like, ‘It'll never happen. Neural nets aren't gonna work. We’re so far away.’ Then suddenly people were doing it because the hardware caught up. So I completely jumped into doing this with audio.”
Sourcing from the Beatles and NOFX for their initial experiments, Dadabots made parody albums using generative technology. “We started to make deep fake albums imitating our favorite bands.” Explaining the basics of the science behind their approach, “Audio is a sequence of amplitudes that make a waveform, so any machine learning model that deals with sequences, including the same architectures that drive text models like ChatGPT can also apply to audio.”
The project was the first of its kind and the albums quickly gained the attention of millions of listeners online. Forming a cult following, fans started making their own satire videos. “It was a really, really fun time experiencing that,” says CJ. “I love the comedy of it all. That's probably the best part in addition to working directly with artists.”
Renowned UK beatboxer Reeps One–the first beatboxer to successfully imitate UK dubstep– caught wind of the project and requested a model be trained on his beatboxing. The collaboration resulted in a breathtaking video filmed in the anechoic chamber at Bell Labs featuring Reeps One battling the Dadabots’ programmed model: Reeps One & Dadabots.
In 2019, Dadabots launched their 24-hour AI-generating death metal livestream, followed by a series of others, and gained further notoriety after receiving press attention from major outlets like VICE, TIME, Bloomberg, The Guardian and more.
Along this journey, CJ was introduced to web3 via hackathons. “I travel the world for them. It's the most fun experience ever,” says CJ. Drawn in by the culture, CJ has now participated in over 70 hackathons. “The culture around open-ended creativity without restrictions is so palpable. The products don't have to be commercial or anything useful. They can just be entertaining, or interesting, or nerdy, or impressive, or avant garde. That kind of creative freedom–with coding and the community that surrounds it–is so inspiring.”
Constantly feeding his own creative well, CJ is now starting to see the fruit of his work’s ability to inspire. In a full circle moment, CJ and Zack recently taught a workshop back at Berklee, where they met a student who’s senior project was to cover a Dadabots song. “It was mind blowing,” says CJ. “He’s trained all the same models that we've trained. We were like, ‘Holy shit, we have to support this kid, he's like 21-years old and already catching up to us.”
Whether pioneering on the fringe with Baroquecore-style, Ethereum-themed epics, and fusions of human and machine learning, or inspiring the next generation to run their own experiments, Dadabots’ historic footprint is only becoming more monumental by the day.
****
Collect Music NFTs from HIGH FREQUENCY Volume 1. “RUG THE BOTS” is the ninth track to be released from NOISE, for HIGH FREQUENCY Volume 1. “RUG THE BOTS” is dropping via Sound Swap on May 04 at 2pm PST.
HIGH FREQUENCY Volume 1 is a compilation album curated by NOISE, with 16 tracks released on Sound. The album features new songs from artists we believe represent the values of web3 and whose work will be essential to every music NFT collector’s set in the future.
Collect and add HIGH FREQUENCY writing NFTs to your digital library. There are 10 editions of this article available for 0.003 ETH on Mirror.
Article by Wallace Morgan for HIGH FREQUENCY, NOISE’s weekly newsletter. Subscribe for more.