Chasing Frisson: Creating AI Music That Gives Me Real Chills

The Frisson Moment

The first time it hit me, my room was dark except for the glow of my desktop computer screen. It was past midnight, the kind of hour when the city outside feels like it belongs to someone else. I had just finished prompting my first track. A love song in English called Oh Loving You. A melody that sounded like it had been pulled from a half remembered dream. I hit play, leaned back, and waited for the usual polite nod of approval my brain usually gave AI output.

Instead, my forearms prickled. A low hum started somewhere behind my sternum and spread outward until my whole body felt like it was vibrating in tune with the bass. That unmistakable rush. Half electricity and half ache. It raced up my spine and settled behind my eyes. I knew the word for it: frisson. The same physical shiver great music has always delivered. Only this time the song had never existed before I asked a machine to make it.

I have been chasing that feeling ever since.

Every week I sit down. I refine prompts the way a songwriter once refined chord progressions. I generate until something inside me answers back. Then I master the vocal timbre. I swap the instrumentation. Sometimes I translate the lyrics into Greek or Spanish just to hear how the same emotional core lands when the phonetics change. The new version often hits harder. Different drums. Different timbre. Same gut punch. I call these my frisson songs. They are the only metric that matters to me.

I upload every finished track through a basic DistroKid account to Spotify, Apple Music, YouTube Music, Amazon, everywhere the service reaches. I do it the same way thousands of other independent creators push their work into the void. Quietly, hopefully, without expectation of virality. So far none of my songs have broken through. No sudden spike in streams. No playlist adds. No late night texts from friends saying this one wrecked me. The numbers stay small. The feeling stays private.

That disconnect is the part most people outside the game never see.

Voices from the Debate

• Traditional musicians often look at this wave and see theft.
  In long raw threads on social platforms they argue that AI users skip the years of scales, the failed gigs, the nights staring at a blank page until something honest appears. They call it laziness dressed up as innovation. One creator summed it up plainly: even an unskilled human musician gets respect if they tried. The AI crowd in this view did not even attempt the work.
• The people actually making the tracks push back with a different story.
  They talk about iteration as craft. Hundreds of generations discarded because the vocal did not land in the chest, because the drop felt empty, because the emotional temperature was off by a single degree. They describe prompting not as absence of effort but as a new form of listening: you learn which words coax warmth from the model, which structures make the hair on your arms rise. One user put it this way during a back and forth that stretched for hours: AI predicts the next note the same way a human songwriter who has absorbed thousands of records does. It is synthesis, not paste. The output has never existed before in exactly that combination. Therefore it cannot be plagiarism in any legal or artistic sense.
  
  The exchange ended the way these conversations usually do. Both sides convinced the other will never understand. One side sees cultural erosion. The other sees a tool that finally lets them capture something they could only chase before.
• What gets lost in the shouting is how many working professionals are already using the same technology quietly.
  Producers whisper about generating rough mock ups in minutes, then hiring live players to re record the parts that matter. Songwriters admit running lyric ideas through models to break writers block, then rewriting everything in their own voice. One estimate floating around music Twitter puts the figure at roughly 60 percent of current mainstream radio material having some AI assistance at the demo stage. The public never hears the machine version; they hear the polished human one. The line between real and fake is already blurrier than the gatekeepers want to admit.

The Scale of the AI Music Explosion

Right now roughly 100000 new tracks enter the global music ecosystem every single day. A growing share of them are fully AI generated. One major platform recently reported that 44 percent of its daily uploads. About 75000 tracks. Are now produced end to end by artificial intelligence. That is more than two million AI songs a month.

The Private Metric and the Future Ahead

This is the part the broader creator economy rarely discusses out loud. Most independent artists whether they stream gameplay, post short videos, or upload songs live in the gap between the story they tell themselves and the metrics the platforms return. You can spend months perfecting a voice, a visual style, a prompt workflow, only to watch the numbers stay flat while someone elses half finished clip explodes. The mental math becomes corrosive. If no one is listening, am I still an artist? I feel that pressure acutely because the external world keeps reminding me my chosen tool is suspect. Yet the internal compass does this make my body respond remains steady.

That private metric may be the most radical thing about this moment. In an attention economy built on public signals likes, streams, watch time a handful of creators are optimizing for something the platforms cannot measure: a physiological reaction that happens in the dark, alone, when no one is watching. It is inefficient. It is unmonetizable in the short term. It also feels to the people experiencing it like the only honest north star left.

Look a little further down the road and the picture shifts again. The same platforms now flooded with AI tracks are quietly evolving. Some already function as marketplaces where pro subscribers can commercially release what they generate. Others are experimenting with features that connect AI generated songs to human singers and live performers who want fresh material without the traditional A and R gatekeeping. The logical next step is integration under one roof: generate, refine, pitch to a network of real artists, split royalties, book live shows. The AI becomes the idea farm; the human becomes the interpreter and the face. Distribution, discovery, and performance all happen inside the same ecosystem instead of scattered across DistroKid, email pitches, and unpaid open mic sessions.

Whether that future arrives in two years or five, the direction feels inevitable. Technology has always redefined what counts as real music. Electric guitars, samplers, Auto Tune, bedroom pop. Each time the old guard predicted the death of craft. Each time craft simply changed shape. The uncomfortable truth is that the barrier to entry has collapsed. Anyone with a laptop and a decent prompt can now make something that gives another human being chills. That fact threatens livelihoods and upends identity for some. For others it removes the last excuse not to try.

I in my dark room do not pretend my tracks will replace anyones favorite band. I do not claim the machine feels anything. I only know that when the right combination of code and iteration lands my body answers. I export the file, upload it to every service I can reach, and go to bed. In the morning I will check the numbers probably still low and then I will start again. Not because the world is asking for more music. Because something inside me is.

And maybe just maybe that quiet insistence is the most human part of the entire story. In a sea of content optimized for algorithms one person is still chasing the moment when the algorithm disappears and the body takes over. The machine helped me find it. The feeling belongs to me. I keep sending it out anyway one upload at a time into a world that mostly scrolls past.

The frisson remains. Everything else is just noise.