I maybe a lot of things, but I’m not a rapper. I discovered it when I was asked to do a few freestyle verses during a visit to the Abbey Road recording studios in London. Immediately, lines from famous rappers come to mind – classic Biggie, a few yelps from Young Thug, the theme from The Fresh Prince of Bel-Air – but I have to think up something original.
In desperation, I decide to rap about my morning routine. Adopting a slow pace and a simple rhyme scheme that even the Sugarhill Gang would disdain, I begin: “I get up at seven and brush my teeth. I’m already lost. What rhymes with “teeth”? Panicked, I stare at the computer in front of me, which is running a demo of iRap, an artificial intelligence software designed to aid in real-time lyric writing. He transcribed my words and suggested possible rhymes I might want to use: “moor, sheath, under”. Could this work? “Make a bacon sandwich, put some cheese underneath,” I sigh. I even fell below my own lowest standards.
iRap is the first product from music startup BrainRap, which aims to combine music and neuroscience. The software uses language processing algorithms to suggest words or phrases based on cadence, rhyme, and meaning. Its creators are an unlikely pair: Micah Brown grew up on the south London grime scene and signed to a Sony-affiliated label before getting into tech, while CJ Carr is a Boston metalhead who worked on projects including algorithmic death metal. Generator.
The pair met at a tech event at the Massachusetts Institute of Technology (MIT) and spent their early hours entangled in a frantic beatbox session. Later, inspired by Brown’s experiences seeing rappers struggle to write lyrics and drawing on Carr’s expertise in machine learning algorithms, they began to prototype what would become iRap. A six-month stint at Abbey Road music tech startup incubator Red helped bring the project to fruition.
Brown was the first black founder to have a company in the program. “There’s a clear connection between Abbey Road and the Beatles, but I also saw Kano play there,” he says. “For me, as a Black Brit, as a South Londoner who didn’t grow up with much, being included was a big deal.”
The iRap software processes lyrical input through multiple layers of technology. The first, text-to-speech, is relatively simple, although the transcription quality is remarkably accurate. Words are then fed through natural language processing, which classifies parts of speech, sounds, and stressed syllables. Then suggestions are created by algorithmic language models that can be trained to give a probability that one word will follow another, as used in the auto-complete feature on smartphones.
Carr’s proprietary algorithms include something he calls Phonetilicious. “It will take a sentence and swap the nouns, verbs and adjectives to maximize potential alliteration while keeping the meaning intact,” he says. “So if I said ‘big red dog,’ it might suggest ‘colossal crimson dog,’ which sounds a lot more musical.”
But would you really want to say “colossal crimson canine” in a rap? Later, I continue to play with the software at home. When I say “big red dog in the house”, he offers me rhymes such as “mouse”, “grouse” and “slaughterhouse”. Although I feel strangely driven to rap about a barnyard massacre, the rhymes themselves are solid. The synonyms, however, are something else: I am offered “adult red goat in the dwelling”, “cosmic coco cow in the enclosure” and “monolithic red man in the mansion”. These alternatives can be nice in terms of alliteration and assonance, but they also sound very silly.
I test a few lines of established rappers to see what the software does with them: Tupac, Future, a bit of Grandmaster Flash. I rap the first line of the classic NY State of Mind by Nas: “Rappers; I spin ’em with the funky beat / I kick, musician inflict the composition. Then come the rhymes: “suspicion”, “acquisition”, “abolition”. The algorithm comes up with synonymous phrases: “the pimps are stuffed with locations and with the musical punishment portfolio of the cowardly chick pacing”; “Homeboys are full of hops and with the funke [sic] heartbeat honey jazz demographic severity.
Besides the fact that these suggestions are military-grade nonsense, they also point to known issues of human bias that creep into algorithmic datasets. For one, the synonyms I’m offered for “rappers” include “thugs,” “pimps,” and “gangbangers,” suggestions that likely reflect the racial biases of the data. It also indicates that the software struggles with colloquial language and, as Brown readily admits, is better at interpreting a standard North American English dialect. This doesn’t bode well for a genre like hip-hop, which relies heavily on slang. Theoretically, the use of machine learning should mean that as more people use the software, it will improve understanding of various languages and accents.
Any announcement of new AI-based music technology inevitably comes with questions about the future of creativity and whether automated tools pose a threat to human artistic expression. Br!dge, a British-Jamaican musician who helped test the software, doesn’t see it that way. “A few years ago we might have had this same conversation about creating music on a computer or using a sampler,” he says. Instead, musicians used these tools to create exciting new sounds that had never existed before. Carr thinks AI composition tools will be commonplace in the future. “They will be a simple part of the music production studio,” he says, “just like a synthesizer.”
I still wonder if professional rappers, who pride themselves on their lyrical dexterity, will welcome this algorithmic intervention in their creative process. Br!dge is also uncertain. “If the target market was rappers, I think they’d be concerned, because the whole point of being a lyricist is having your own thinking and having your own mind,” he says. “I probably wouldn’t use iRap for my own artistry because I consider myself independent, a purist, someone who takes pride in designing my own lyrics.”
In this case, who is iRap really aimed at? Besides recording his own songs, Br!dge also writes lyrics for other musicians. He thinks songwriters working in the commercial music world could greatly benefit from the software. “There are times when you’re just trying to get something going and get your front lines quickly,” he says. “The song isn’t necessarily personal to you, so it just helps you produce as many high-quality songs as possible to deliver to different labels.” It might sound like an unromantic approach, but we live in a time when it’s not uncommon for a pop hit to have more than 10 credited songwriters. Carr and Brown also suggest another audience: people like me, beginners who want to learn the ropes of timing, rhyme and flow. “If you’re not an expert, you can lose your rapping momentum and find yourself in a bind,” Carr says. “But you may look at the screen and realize that there are four directions you can go right now. You can say things that sound phonetically similar, talk about related concepts, or just keep the momentum going so you don’t stumble. on yourself.
The more I test iRap, the more useful I find it to have my words transcribed and offered rhymes, even if not all suggestions are viable. The software is like training wheels, helping me relax and improvise, letting me know I have support if I need it. Carr suggests hobbyists could use the software when they’re just starting out and then grow past it. “If tools like iRap are well designed, they’re not crutches,” he says. “They give you a skill and then you can uninstall the tool and you still have those abilities.”
After spending a few days rapping on my computer, I still think it’s unlikely I’ll try to perform in public anytime soon. But thanks to the algorithmic training, I’m confident that the next time I’m put in the hot seat, I can at least make it to the second line.