This El Paso startup wants to become 'the ChatGPT for music' – The Business Journals
There are thousands of aspiring musicians with dreams of writing and recording songs to share on social media or other platforms. Samuel Moreno counted himself among that group.
The El Paso native played the guitar in high school and in college at Missouri’s Park University. While he joined several bands in high school, Moreno wanted to spin that playing into a full-time career as a guitarist.
But Moreno also wanted to make a more tangible impact on the world, something he wasn’t sure being a guitarist would bring.
“When I graduated from high school I had to make that decision if I wanted to continue pursuing music or if I wanted to go into computer science,” Moreno told Albuquerque Business First. “So I made that decision to go to computer science, but essentially what I really wanted to do was play music.”
His computer science studies at Park University helped Moreno land a remote job at Oculus, the virtual reality startup bought by Meta (then Facebook) that was best known for its Oculus Rift headset product. Moreno said his job at Oculus was demoing that product for potential customers, clients and investors.
The founder then moved to Phoenix while working at Cruise, the San Francisco-based autonomous driving startup. Moreno stayed there for eight months before starting at Imagry, another startup working on autonomous driving technology.
“That’s when I really got my grasp into artificial intelligence and the use cases and what it can do,” Moreno said.
After Moreno brought his guitar into the Imagry office one day, a machine learning engineer at the startup mentioned an idea — “Why don’t you train a machine learning model to write a song for you?” A few months later, that idea would turn into the very earliest edition of MusicStar.AI, the startup Moreno co-founded earlier this year.
He and co-founder and chief technology officer Kinyugo Maina, who lives in Kenya, released MusicStar.AI’s first minimum viable product (MVP) — a software application that Moreno said has accumulated over 125,000 users — in January 2023. The pair want the startup to become “the ChatGPT for music,” he said.
Moreno and Maina are currently doing that by building algorithms to train machine-learning models on a diverse set of copyright-free and unlicensed musical data. People can then type in prompts that those models would use to generate lyrics and background beats and instrumentals for their own music.
MusicStar already has some competition. Companies like Soundful and Splice AI have already raised millions toward their artificial intelligence-backed music-generating offerings.
But MusicStar could be on the fast-track to similar valuations. Moreno was tapped for the ACT Tulsa Accelerator this spring alongside nine other companies out of 225 applicants. The six-month accelerator ends with a demo day on Oct. 5.
Moreno said the mentorship offered through the accelerator has been helpful for MusicStar’s growth. There aren’t a lot of examples of other companies working in AI to “mimic,” he said.
After raising $100,000 in a friends and family round and landing $70,000 thanks to its acceptance into the accelerator, Moreno said he’s looking to raise an additional $250,000 for the upcoming demo day. He and Maina, alongside three contract developers, are currently building MusicStar’s second product.
That product would be a big jump over the startup’s first MVP, Moreno said. Instead of a very basic text-to-music model, he said the startup is working on incorporating a 3D avatar into the application while expanding MusicStar’s initial text-to-music model.
To access more “data” for that model in the form of songs, Moreno said the startup has kicked off a pilot program with a U.K.-based record label that will give MusicStar access to that label’s 20,000-song catalog.
That’s the biggest step forward for the startup — striking partnerships with other record labels to access licensed songs in order to train more sophisticated models. More data via licensed music, plus building better algorithms to use that data more efficiently, will be how MusicStar grows the quality of its AI songwriting tool, Moreno said.
And it wants to do that from El Paso — not Silicon Valley, the global tech hub that has become synonymous with generative AI and machine learning startup development.
“I really wanted to build the startup here,” Moreno said. “El Paso is closer to my family and friends, and of course, it’s a lot cheaper than being in Silicon Valley.
“Right now, you could build anywhere in the world. You don’t have to be in Silicon Valley to build the next ‘unicorn,’ how they say out there,” he continued. “I wanted to put that I came back because it’s a lot slower-paced environment compared to being out there.”
Moreno said the startup has worked with local organizations, like El Paso’s STEE Foundation and a sprint program at New Mexico State University’s Arrowhead Center, to boost its development.
Want to stay ahead of who & what is next? Sent weekly, the Beat is your definitive look at New Mexico’s innovation economy, offering news, analysis & more on the people, companies & ideas driving your city forward. Follow The Beat