How I Got Here: An Old-School Jazz Singer Becomes a Liminalist

For more than twenty years, I worked as a vocalist, songwriter, and collaborator within the jazz community in Denver. Music was never simply a performance activity for me — it was a way of thinking, listening, improvising, connecting, and understanding the world.‍

Leaving the City
Just before the pandemic, I made the decision to leave the city and move to a small mountain town an hour’s drive from Denver. It was a choice rooted in quality of life: more quiet, more nature, more space to breathe. But like many meaningful choices, it came with tradeoffs I didn’t fully understand at the time. 

Maintaining a life in music became harder. 
The jazz community I had worked with slowly dispersed. The pandemic accelerated a trend that was already beginning to happen. Musicians moved away. Venues changed. The culture surrounding live music shifted dramatically. A prime example: In December 2020, El Chapultepec closed. After 87 years of operation (since 1933) this famous venue shut its doors for good due to economic struggles exacerbated by the COVID-19 pandemic and changing neighborhood re-development.  

Music, however, remained one constant in my life. I found myself still writing songs, still hearing arrangements in my head, still wanting deeply to create. At the same time, my relationship to live performance began to change. Gone were the days when I was willing to sing in bars while televisions flickered overhead and I had to ask the staff to turn the football game off before we started playing. Increasingly, audiences did not seem willing to truly listen to music anymore. They wanted music as a backdrop — wallpaper — which can be deeply discouraging for serious musicians whose work depends on attention, nuance, improvisation, and emotional presence.  

What emerged over those years was a growing sense of isolation. Musicians I cared about and respected were now living in different states. I was behind the learning curve in collaborating remotely. Musicians were sharing files, meeting over Zoom or telephone, talking through arrangements and ideas from hundreds—or even thousands of miles away. I began to understand that the future of creative work for me would depend upon learning how to use technology at a much deeper level.  

10 grand on my first CD in the early 2000s!
When I look back on the first CD I ever produced, I find it almost surreal that I spent close to ten thousand dollars on that project: hiring top musicians, booking excellent recording studios and engineers, commissioning artwork for the jewel case — in essence, navigating all the costs associated with professional music production at the time. Remember CD Baby anyone? Much of that old model has become financially prohibitive for independent artists, even those who spent years balancing music alongside professional careers as I did. Continuing to make music affordably would require me learning how to do many new things myself.   

Unexpectedly, that became exciting.
I built a bare bones recording studio in my home. I bought a Rodes condenser mic. I learned how to set levels on a Focusrite Scarlet audio interface. I learned to send and receive audio files from other musicians. Dropbox was surprisingly hard for me to figure out for some reason. From there, I began learning recording software like Audacity and GarageBand. I learned how to edit tracks, though setting EQ levels and fiddling with compressors is not my strong suit. That essential part of the process is still best left to people who have a knack for it. I learned how to distribute music through TuneCore and LANDR, and release work on platforms like Spotify. What initially felt overwhelming gradually opened an entirely new creative landscape for me. The learning curve itself became part of the artistic experience.  

Then AI arrived. 
Like many musicians, I initially approached AI with skepticism and concern. Some of the most gifted musicians I know reacted strongly against it for understandable reasons. If have dedicated your life to an instrument, then the idea that artificial intelligence could somehow replace human artists is deeply unsettling — even devastating — to contemplate. I think the deepest feeling of loss is tangled up with the fear that music itself will no longer be valued as a profoundly human form of expression.  But over time, I began to see AI differently. Not as a replacement for musicians, but as another evolving tool: imperfect, provocative, sometimes surprisingly useful, and capable of expanding experimentation in ways that reminded me of other technological shifts musicians have already adapted to over the decades.  

As a songwriter — my strength has always been lyrics, melody lines, and vocals — this became especially interesting to me. Even in my earlier recording projects, I was always collaborating with arrangers and instrumentalists who helped shape the final sound of the music I wrote. There were many times when another musician arranged one of my songs so that it sounded very different from what I had originally heard in my head; there was often no easy way to bridge that gap. Part of being a songwriter meant accepting that by the end of the collaborative process, the song no longer belonged entirely to me.  

What I discovered with AI music tools was something completely different. True, I was still giving my lyrics and arrangement ideas over to a collaborator— in this case to an AI tool. But what was entirely new was the number of choices I could actually make through trial-and-error use of style settings, voices (including some loosely based on my own voice samples), and many other settings. I returned to songs I had written twenty years earlier and suddenly heard them interpreted in entirely new ways. The same lyrics could expand across genres, moods, tempos, and sonic textures I never would have been able to explore before. I could try out a Toots Thielemans-style harmonica line, a Béla Fleck-inspired banjo, a sparse acoustic arrangement, or a groove-driven rhythm section simply to see what happened. In another era, that kind of experimentation would have required musicians, studio time, and a much larger budget.

For me, that opened an entirely new level of creative play.  The technology is imperfect, of course. Sometimes Ai produces material that is genuinely terrible. More often the results are shallow, strange, or unusable, and I abandon the attempt entirely and return to the drawing board. As a vocalist, I am quickly bored by most AI-generated vocal options. The voices often sound like simplified composites of contemporary pop singers—recognizable enough to feel familiar, but lacking the individuality that makes a voice memorable. There is still no voice-generation technology that I know of capable of truly replicating the nuances of my own vocal style or phrasing in a way that feels emotionally authentic to me. Because of that, I often find myself splitting tracks, rendering instrumental stems and replacing AI-generated vocals with my own recorded vocals. This process can be surprisingly difficult and technically complex. I have to admit that the AI vocals often sound better alongside the AI instrumentals. What does work fairly well is to add back up singing tracks to the AI lead; a bit of vocalese harmony sounds nice on bossa nova tracks especially. 

I have wondered whether part of the difficulty in singing to an AI track comes from the way AI-generated music itself is “formed.” AI-generated music occupies a curious middle ground. It generally operates within the familiar twelve-tone framework of Western music, yet something about it often feels slightly unstable. The pitch may be technically correct, the harmonies recognizable, and the rhythms coherent, but the music can still seem oddly “squishy” around the edges—as though it is approximating musical intention rather than fully inhabiting it. The effect becomes especially noticeable when the track is separated into stems. Stripped of the masking effect of the full mix, individual parts sometimes reveal strange pitch relationships and ghostly artifacts that feel neither fully intentional nor entirely accidental. What I have read about this confirms my experience. Working as a vocalist alongside great musicians feels like standing on solid ground.The music can twist and turn in surprising ways, but you know you’ll cross the bar line together. Singing with AI-generated tracks can feel more like standing on a too-soft mattress. The surface supports you, even as it is giving away just ever so slightly beneath your feet. Add to that the fact that a computer-generated vocalist can cram a lot of words into a phrase seamlessly—they don’t need to breathe! For a human, keeping up requires acrobatic articulation. Some of my own original lyrics become a challenge to sing.

What interests me most, however, is that even these imperfections become part of the artistic conversation. The friction itself can lead to new interpretations, new phrasing choices, new arrangements, and discoveries that might never have emerged otherwise.  The key distinction, for me, is this:  The artist’s dream must remain human. The listening remains human. The emotional intelligence remains human. The improvisation, judgment, humor, restraint, memory, risk-taking, and soul remain human.  AI can generate patterns and possibilities, but meaning still emerges through people collaborating, responding, refining, rejecting, and discovering together. And working with Suno will never be as rewarding as working live with the accomplished musicians I admire. A shout-out to guitarists and arrangers Mike Sunjka, Neil Haverstick, and Mark Caldwell; pianists Doug Roche, Trip Ziegler, Eric Gunnison, Hank Troy, Art Lande, and my early mentor, the late Ellyn Rucker; bassists Peter Huffaker and Mary Stribling; drummers and percussionists Thomas Blomster and Ernie Crews; and studio engineer Gary Flori—also a fine drummer. This is only a short list. There are many more musicians with whom I shared stages, studios, and recording projects over the years.

My thanks to you all! —M.