Featured Photo by Isabella Gutierrez
Story by Isabella Gutierrez
At 4 a.m. on Friday, September 15, Evan Scandrett was tired. He sat in a dark soundproof recording studio, padding covering the walls. The only noises he could hear were the squeak as he swiveled in his chair, the vibration of the bass when music oozed from overhead speakers and the clicking of his fingers on his computer keyboard. The ideas were flowing as Scandrett, a Middle Tennessee State University senior audio production major, and his co-writer, a computer algorithm named ChatGPT-4 worked overtime.
This was Scandrett’s first time using artificial intelligence to help him craft a tune. So far, he was impressed.
Scandrett gave his co-writer instructions through prompts. He was looking for a specific sound using combinations of music that he wanted to influence the final product.
He asked the AI bot specific questions. Like, “Can you create a new song in a similar expression but not the same tangible idea. However, I want you to implement the same themes and archetypes.”
And, oh yeah, the song must be set in El Paso, he told the computer.
The reply was instantaneous. “Certainly, I can create a song that captures similar themes and archetypes, with the setting in El Paso.” This was followed by song lyrics to fit Scandrett’s request.
As the sun rose over the horizon, the human in this song-writing duo began “arguing” with his co-writer, going back and forth to ensure none of the new work was copyrighted, a boundary the human songwriter didn’t want to cross. Around 7 a.m. a new song, “El Pasado de El Paso,” was recorded.
ChatGPT-4 uses AI technology to create anything from a morning gym routine to an apology email. Many people know of just ChatGPT but ChatGPT-4 “is more advanced, and beats ChatGPT in just about every category,” said Drew Robb, a columnist for eWeek magazine.
Scandrett had already programmed his ChatGPT-4 by telling the bot he identified as an “audio engineering ethnomusicologist” and for his helper to “channel the spirits of every character mentioned on the ‘Sgt. Pepper Lonely Hearts Club Band’ album art.”
He asked the AI bot to consider the work of David Bowie, Bob Dylan and John Lennon to create a new type of sound but, as he instructed his non-human collaborator, the new work should stay “in the realm of old school folk music to keep it traditional melodies.”
The student songwriter compared the information he gives his cowriter this way: “It’s kind of like a really in-depth Tinder bio so it knows how to respond to you exactly.”
Scandrett is not the first in the MTSU recording industry program to recognize the power of AI use in music.
This past June, Joseph Akins, a recording industry professor at MTSU, also had an experience with AI songwriting. He attended the CD Baby Conference where well-known music business professional Rick Barker presented an AI session. He entered a prompt into ChatGPT, asking it to write a rap with a specific topic and style. Quickly, the rap was finished after only pushing the enter button. Barker played a beat, and asked “can anyone in the audience rap?” A volunteer was invited onstage from the audience and flawlessly performed the rap. It seemed like magic. “You see the text, but when the rapper raps it, it’s just like ‘wow’ it really sounds like a rap, and everyone was really excited,” Akins said. Before the audience’s eyes, the song was finished.
AI is also entering the artist marketing and audio production industries. Last spring, Akins took an online course for AI marketing and learned how to use ChatGPT to create social media posts and newsletters.
“You’re saving yourself time… supposably…from having to write all that out,” Akins said of the benefits of using a bot to promote music. He also mentioned its usefulness in creating album art, graphics and videos.
A few months ago, Akins saw an ad from the music distribution company, TuneCore. In the ad, it mentioned a collaboration between the company and the popular music artist, Grimes. TuneCore has employed AI programming to put Grimes’ vocals onto any song that is distributed through the company. TuneCore’s website states: “Our aim with this pilot partnership is to create a path for artists to consent to have AI versions of their voices used in other creators’ music.” With this partnership, GrimesAI receives 50% of the revenue, this allows new music producers to have a popular singer on their track and hopefully attract more listeners.
Another MTSU Audio Production major, Graham Broom, noticed an intriguing ad for eMaster. This an app that can master your music with AI at an affordable rate, and Broom was eager to try it. He quickly ran a song he was working on for the band, The Broomexstix, through the program. He was impressed with the results. According to eMaster’s website, the app is “An online mastering engine that’s fast, easy to use, and sounds incredible…made by Grammy-winning engineers, powered by AI.” It costs $13 a month and has a free trial for your first song. This product is more cost-effective compared to traditional mastering.
Even though this product amazed the recording industry major, it wasn’t perfect. Graham said, “I’d rather not use it personally. It’s good but you don’t have much control over what you end up with.” However, he did mention the usefulness of this app for independent artists who can’t afford hands-on mastering. “I wouldn’t be surprised if actual hands-on mastering becomes common only for signed and well-funded artists,” he added.
Scandrett hasn’t used ChatGPT-4 for any more songwriting though he is experimenting with the software. He asked the bot to rewrite the first page of “Don Quixote.” He asked the bot to “rewrite it like Roald Dahl” and the bot replied “Certainly, here’s a whimsical reinterpretation of the passage in the style of Roald Dahl,” followed by the passage perfectly written. Scandrett asked the bot to reimagine the story with other authors including the King James Bible, T.S. Eliot and Arthur Rimbaud. Scandrett enjoyed using the program to write his song and as a learning tool to help develop his skills and understand the work of different artists and creators.
Some detractors say using Artificial Intelligence for songwriting takes the writer’s authenticity from their music.
MTSU audio production major, Jeremiah Andrews isn’t concerned about this argument. “I’m fearful of AI but only to an extent. I know a robot won’t be able to convey the same emotions and feelings as a human could in their art.” Scandrett also holds hope that human brain power will prevail.
“I don’t think AI will stop true songwriters from being heard because they’ll always find a way to sound more unique than everything going on in a world full of AI.”
Isabella Gutierrez is a contributing writer for MTSU Sidelines.
To contact News Editor Kailee Shores, Assistant News Editor Alyssa Williams and Assistant News Editor Zoe Naylor, email [email protected].
For more news, visit www.mtsusidelines.com, or follow us on Facebook at MTSU Sidelines or on Twitter and Instagram at @mtsusidelines.
