I Built a Fake Music Artist Using AI - Here's Exactly How

His name is Cormac Riley. He's from Georgetown, Texas. He makes warm, hopeful singer-songwriter music that sounds like Jake Scott and Ben Rector had a musical lovechild.

He doesn't exist.

I created Cormac Riley - his name, his sound, his entire artistic identity - using a framework I developed for building consistent AI music artists through Suno. And if I played you his songs without context, you wouldn't guess they were AI-generated.

Here's exactly how I did it, step by step.

Step 1: I Picked My Ingredients

Every great artist is a blend. Nobody emerges fully original. Taylor Swift is country storytelling meets pop hooks meets confessional songwriting. The Weeknd is R&B meets dark electronic meets Michael Jackson's vocal influence.

I wanted a singer-songwriter. My reference blend: Jake Scott's vocal warmth and optimism, Ben Rector's piano-touched emotional earnestness, Matt Kearney's rhythmic sensibility and hip-hop-adjacent flow, and just a touch of John Mayer's guitar craft.

Three to five reference artists is the sweet spot. Fewer than three and your AI artist sounds too much like one real person. More than five and the identity gets muddled.

Step 2: I Researched Each Artist Like a Producer Would

I can't play an instrument. Can't read sheet music. Can't tell you what key a song is in. None of that matters.

I used Claude to research every reference artist through reviews, interviews, and fan descriptions. The goal: extract specific descriptors for how each artist sounds.

Jake Scott: "warm, optimistic, clean vocals, acoustic guitar foundation, modern pop production, hopeful without being cheesy." Ben Rector: "piano-forward, orchestral tendencies, earnest delivery, themes of nostalgia and gratitude." Matt Kearney: "hip-hop influenced, rhythmic vocal delivery, layered production, journey and struggle themes."

Those specific words - warm, optimistic, earnest, rhythmic, layered - become the building blocks.

Step 3: I Created the Sound DNA

From the research, I built what I call a Sound DNA profile. It's a structured document that captures every sonic characteristic of the artist:

Vocals: Warm male tenor, clean and clear, occasional falsetto on emotional peaks, conversational delivery that shifts to earnest belting in choruses. Never gritty. Never dark.

Instrumentation: Acoustic guitar is the foundation. Piano appears as an accent, never the lead. Light percussion - finger snaps, subtle brushed drums. Electric guitar only on upbeat tracks, kept restrained.

Production: Modern but not over-produced. Intimate feel, like a studio session you're sitting in on. Spacious mix - every instrument has room to breathe.

Tempo range: 70-120 BPM depending on the song type. Ballads at 70-85. Mid-tempo at 90-105. Upbeat at 108-120.

Step 4: I Built Prompt Templates

From the Sound DNA, I created three prompt templates - preset configurations for Suno that reliably produce Cormac Riley's sound.

His upbeat template: "Indie folk singer-songwriter, warm male vocals, acoustic guitar driven, light percussion, uplifting and hopeful, modern production, clear and intimate vocal recording, gentle piano accents, 110-120 BPM."

His ballad template: "Acoustic ballad, warm emotional male vocals, fingerpicked guitar, soft piano, intimate and reflective, minimal production, spacious mix, strings in chorus, 70-85 BPM."

His rhythmic template: "Indie pop singer-songwriter, warm male vocals, acoustic guitar with light electric, driving percussion, optimistic energy, polished production, claps and stomps, building arrangement, 100-115 BPM."

Three templates. Three different song types. All unmistakably the same artist.

Step 5: I Wrote Lyrics in His Voice

Cormac Riley writes about family, growth, finding the extraordinary in ordinary moments. He uses second person sometimes - talking to his kids, his wife, his younger self. His metaphors come from nature and Texas landscapes. He doesn't preach. He shows.

I wrote a song about my daughter learning to ride a bike. The chorus: "Look at you go / Wind in your hair and the whole world below / Every wobble every scraped up knee / Was just the road to being free."

That's a Cormac Riley lyric. Direct. Emotional. Specific enough to be personal, universal enough to resonate. And structured with consistent syllable counts and rhyme patterns because that's what makes Suno sing it well.

Step 6: I Generated and Iterated

First generation: vocals were perfect. Guitar was too prominent. Drums felt heavy.

I adjusted one thing - added "minimal percussion, guitar-forward" to the prompt. Generated again. Better. Third try, I added "spacious mix" and got exactly what I heard in my head.

Three tries. Maybe four minutes of total generation time. The result: a song that sounds like it belongs on a Spotify singer-songwriter playlist.

What Cormac Riley Sounds Like Now

After building his profile, I've created over 30 songs under this artist. They span upbeat road-trip anthems, quiet ballads about watching my kids grow up, and mid-tempo tracks about the strange experience of building a life in a small Texas town.

They all sound like the same person made them. That's the part that impresses people. Not that one song sounds good - that every song sounds like a coherent artistic identity.

I've played these for people without telling them it's AI. The reaction is always surprise when I reveal it. Not "oh I can tell" surprise. "Wait, seriously?" surprise.

You Can Build Your Own

Cormac Riley took me about two hours to fully develop - research, Sound DNA, prompt templates, test songs. Now that the system is dialed, a new song takes maybe 15 minutes from concept to finished track.

The same process works for any genre. I've built a pop-rock artist, a kids music project, and I'm working on a country artist. The framework is identical. The genre-specific details change, but the system transfers.

If you want to try the basic version: pick three artists you love, research how people describe their sound, blend those descriptors into a Suno prompt using the formula (Genre + Vocals + Instruments + Production + Mood + Tempo), and generate five test songs. You'll be shocked at the quality difference compared to your random generations.

Cormac Riley will never do a live show. But his songs are real - and they make people feel something. If you want the full framework for building AI artists, including the Claude AI skill files that automate most of this process, I'm packaging everything into a course. The course is live now at ideatomusic.com/course.