“ChatGPT, I have a wild idea. What if I could get Treblo to just read my entire blog and make a song out of it? That’d be interesting.”
That was the question.
Not help me write a song. Not write a song about my life. Not even write a song about the things I like.
I wanted to know what would happen if an artificial intelligence read everything I’d written on my website, formed its own impression of the person behind those words, and translated that impression into music.
Would I recognize myself?
I thought it would be an interesting experiment.
As usual, I had no idea how much work my little question was going to create.
First, We Had to Read Everything
My website, Jindai’s Jumbled Joint, has been around in one form or another since the late 1990s. It started as a mostly empty personal webpage with some graphics, music, and a joke about the letter J and the number five. (That needs some explaining. The original site was Jindai’s Just Another Jumbled Joint. Five words, see, and a lot of J’s. Now that I feel more unique, I’m not “Just Another.”)
Since then, I’ve written about just about anything that happened to catch my attention.
Movies. Television. Food. Writing. Technology. Jobs. Memories. Music. Football. Personal experiences. Occasionally something serious. Frequently something ridiculous.
Sometimes a simple question becomes an investigation that takes considerably longer than anyone could reasonably justify.
Anyway, ChatGPT initially wanted to approach the project by identifying a few representative articles.
Nope.
That wasn’t the experiment.
I wanted the entire published collection considered. Every article was a mile marker. Something that seemed trivial today might have represented an important moment when I wrote it.
So we went through the archive chronologically.
There were sixty published WordPress entries, including several pages that served more as organizational structures than conventional articles. The earliest eight posts had originally appeared on my LiveJournal account between 2010 and 2012.
We also examined the original late-1990s website, which I’d reconstructed earlier this year using material recovered from the Internet Archive.
Along the way, I supplied historical context that wasn’t necessarily evident in the writing itself.
For example, an article about experimenting with breakfast in July 2020 might look like nothing more than a little cooking adventure. But I’d spent a considerable period homeless before moving into a rooming house in September 2018. Nearly two years later, I had internet access, an air fryer, and enough stability to start writing about something as ordinary as breakfast.
That article meant something different in that context.
We also discussed the disappearance and eventual restoration of my website, the history behind some of my fiction, and the difference between when something was written and when it was published.
All of that took some time—more than expected, to be honest. And I felt I had to add clarifying directions throughout, not to steer the output, but to provide some context I thought might matter.
And then ChatGPT started talking about how we should decide what the song was going to say.
Wait a minute.
“No, You’re Missing the Point”
I had to correct ChatGPT. I wasn’t asking to collaborate on an autobiographical song. I wasn’t going to choose the theme, select the important experiences, or decide what emotional message the lyrics should convey.
In fact, I wanted to keep my opinions out of the creative process entirely. The whole point was to see what an external intelligence would make of my writing. If I helped decide what the song should say, I would already know what I was supposed to hear.
That wouldn’t be nearly as interesting.
Eventually, ChatGPT understood.
It would interpret the material independently, create the lyrics and musical instructions, and let Treblo turn those instructions into recordings.
I would listen.
That was my entire job.
Experiment One: ChatGPT Writes, Treblo Performs
For the first attempt, ChatGPT wrote a complete set of lyrics titled A Little Corner of the Noise.
🎵 Click to read the lyrics — A Little Corner of the Noise/Up All Night.
[Intro — spoken, over a faint electronic melody]
Brought to you by the letter J
And the number five.
[Verse 1]
I built a little room inside a wire,
Hung a couple spotlights, called it mine.
Didn’t have a sermon or a grand design,
Just a borrowed tune and a little time.
The world was getting louder by the minute,
Everybody had a place to be.
I put a sign above an empty doorway,
And wondered who might wander in to see.
[Verse 2]
I could take a thought and chase it sideways,
Follow every branch until it bent.
Turn a little question into twenty,
Then forget the place I meant to went—
Well, that doesn’t rhyme, but neither do I,
And I’ve never let that stop me yet.
There’s a tune that’s playing somewhere in my head,
And I haven’t found the off switch yet.
[Chorus]
There’s a little corner of the noise
Where the broken things still make a sound,
Where a question turns into a story
And a story turns the whole thing ’round.
I don’t always know what I’m looking for,
I just know when something catches light.
If the world won’t give me an explanation,
I’ll be up all night.
[Verse 3]
Some years went missing in the middle,
Some doors closed without a sound.
I learned how little you can carry
When you’re trying not to hit the ground.
Then one day I was making breakfast,
Had a place to sit, a thing to try.
It wasn’t much to write a story about,
But I wrote it down, and I know why.
[Verse 4]
Then somebody pulled the plug on yesterday,
And my little corner disappeared.
I went digging through the ghosts of old machines
For the words I’d left across the years.
Found a page that still remembered music,
Found a joke that hadn’t learned to die.
Put the pieces back where I could find them,
Left the cracks where I could see the light.
[Chorus]
There’s a little corner of the noise
Where the broken things still make a sound,
Where a question turns into a story
And a story turns the whole thing ’round.
I don’t always know what I’m looking for,
I just know when something catches light.
If the world won’t give me an explanation,
I’ll be up all night.
[Bridge — quieter, reflective]
Someone asked me how I’d end the story.
I was young enough to think I knew.
Now I’ve got a thousand little endings
And a thousand things I haven’t gotten to.
I’ve been wrong about a hundred things,
And I’ll probably be wrong again.
But I’d rather leave the question open
Than pretend I knew the answer then.
[Instrumental break — playful electric guitar and keyboard trading phrases]
[Verse 5]
Now there’s movies waiting for an argument,
And a map that took a week to draw.
There’s a story hiding in a deadline,
And a dragon with a bone to gnaw.
There’s a voice that asks me what I’m doing,
And I tell it, “That’s a fair complaint.”
I was only going to check one little thing—
Now the printer’s out of ink again.
[Final Chorus — full band]
There’s a little corner of the noise
Where the broken things still make a sound,
Where a question turns into a story
And a story turns the whole thing ’round.
I don’t always know what I’m looking for,
And I don’t need to get it right.
If the world won’t give me an explanation,
I’ll be up all night.
[Outro — music gradually strips back to the opening electronic melody]
I built a little room inside a wire.
Didn’t know what I would put inside.
[Spoken, amused]
Well… that got out of hand.
It also supplied musical instructions calling for alternative rock, power pop, folk rock, and some nostalgic electronic elements.
I pasted the lyrics and instructions into Treblo.
I didn’t choose a title.
Treblo took care of that.
It generated two versions, which it named Corner of the Noise and Up All Night.
Corner of the Noise
Up All Night
I liked both.
Corner of the Noise was my favorite, although Up All Night had some good points.
The titles were interesting, too. I’ve certainly spent a great deal of my life staying up all night. Corner of the Noise was more abstract, but I didn’t mind it.
There was one thing about both recordings, though. They sounded rather like the 1980s. In fact, they reminded me somewhat of Tiffany’s I Think We’re Alone Now.
Now, I certainly lived through the 1980s. That’s when a lot of my early adult experiences happened. But it’s 2026. I don’t necessarily think of myself as an ’80s guy.
And I wondered whether the sound had more to do with the musical instructions ChatGPT supplied than with anything the website actually suggested.
Which led to another question.
What if we let Treblo write the lyrics, too?
Experiment Two: Let Treblo Decide
This time, I asked ChatGPT to create a description of the person and writing patterns it had identified from the website.
But instead of writing lyrics, ChatGPT would give Treblo that description and let it make all the creative decisions.
Genre. Instrumentation. Vocal style. Lyrics. Structure. Title.
Everything.
The description discussed curiosity, humor, elaborate investigations, unfinished projects, personal difficulties, creative work, mistakes, and the tendency to follow an unexpected question wherever it might lead.
It didn’t prescribe a musical era. And it didn’t tell Treblo what conclusion to reach. I pasted the description into Treblo, leaving the custom lyrics field empty.
Once again, Treblo generated two songs.
This time, it called them Keep Pulling the Thread and Mistaken Not Dishonest.
Keep Pulling the Thread
Mistaken Not Dishonest
Both versions used essentially the same lyrics, but the musical arrangements were different.
🎵 Click to read the lyrics — Keep Pulling the Thread / Mistaken Not Dishonest
[Verse 1]
Started with a room too big and nothing on the walls
I typed a little nothing just to hear it echo back
Then a hamburger became a footnote
Then the footnote became a map
The world kept folding into itself
A whistle a wrinkle a misplaced fact
[Pre-Chorus]
I lost the plot I found it in a drawer
I lost the site I found a scar
Everything I buried I came back for
Everything I buried I came back for
[Chorus]
Keep the thread keep the thread keep pulling it
Wrong turns count as evidence
Keep the thread keep the thread keep pulling it
I was mistaken not dishonest
So I keep pulling it
[Post-Chorus]
Oh-oh oh-oh
Pulling it
Oh-oh oh-oh
Pulling it
[Verse 2]
I chased a melody through thirty borrowed years
Got distracted by a question nobody asked
Wrote a whole investigation
To avoid a single task
And the map I drew of someone else’s season
Was the only thing that lasted
[Pre-Chorus]
I lost the plot I found it in a drawer
I lost the site I found a scar
Everything I buried I came back for
Everything I buried I came back for
[Chorus]
Keep the thread keep the thread keep pulling it
Wrong turns count as evidence
Keep the thread keep the thread keep pulling it
I was mistaken not dishonest
So I keep pulling it
[Bridge]
Half a home half a habit half a hard reset
I rebuilt the ordinary with both hands
You can call it wasted you can call it unread
I call it the best part of the plan
[Chorus]
Keep the thread keep the thread keep pulling it
Wrong turns count as evidence
Keep the thread keep the thread keep pulling it
I was mistaken not dishonest
So I keep pulling it
[Post-Chorus]
Oh-oh oh-oh
Pulling it
Oh-oh oh-oh
Pulling it
And the second one got to me. I mean that literally. I had tears in my eyes. Same words in both, but the music itself added something for me.
Wait. Where Did That Come From?
There were two particular passages that caught my attention.
The first was:
I lost the plot, I found it in a drawer.
I lost the site, I found a scar.
Now, those are wonderfully allusive lines.
Someone listening to them might think about a lost relationship, a forgotten ambition, or something from their past that they’d rather not revisit.
There’s room for interpretation. But I knew exactly what they meant to me. My website vanished. And it hurt.
I’d spent years writing things, collecting memories, and building a little corner of the internet that belonged to me.
Then it was gone.
Eventually, I recovered much of it. But recovering something doesn’t necessarily erase what it felt like to lose it. That word scar was particularly effective.
I hadn’t asked Treblo to describe the loss that way. It came up with that imagery itself.
And then there was another passage.
Got distracted by a question nobody asked.
Wrote a whole investigation.
To avoid a single task.
Oh, come on.
That’s ADHD without ever mentioning ADHD!
And it is so very, very familiar.
There’s something particularly amusing about spending hours researching an interesting question that nobody asked, producing something elaborate and possibly useful, and then remembering that you were supposed to be doing something entirely different.
I’ve done that. A LOT.
Here’s the interesting part: ChatGPT’s prompt never actually used the term ADHD. It described the behavior, but didn’t label it.
Treblo took that description and turned it into three lines that I immediately recognized.
The prompt did mention losing access to years of online writing and recovering some of that history. But it didn’t tell Treblo how that experience felt to me.
And it certainly didn’t instruct the system to use the image of a scar.
Those were its own lyrical decisions.
Mistaken, Not Dishonest
There’s another line that appears throughout the song:
I was mistaken, not dishonest.
That comes from a distinction I’ve discussed in my writing.
Being wrong isn’t the same thing as lying. I can believe something, repeat it in good faith, and later discover that I was mistaken.
That’s different from knowingly trying to deceive someone.
Apparently, Treblo thought that distinction was important enough to make it part of the chorus. I found that interesting.
It also came up with another line:
Wrong turns count as evidence.
I rather like that.
Not every investigation ends where I expected it to. Sometimes the detour turns out to be more interesting than the original question. Sometimes it turns into an entirely new project. Sometimes it turns into an article.
And occasionally, apparently, it turns into four songs.
So, Did It Work?
I think it did.
Not because the songs provide some definitive psychological portrait of me. They don’t. And not because an artificial intelligence has somehow discovered the secret meaning of my life.
It hasn’t.
What happened was considerably more interesting than that.
I supplied a body of writing. ChatGPT examined it and translated its observations into a description. Treblo interpreted that description and produced lyrics and music.
And when I listened, I recognized things that mattered to me. Some were funny. Some were surprisingly accurate.
And a couple were emotional enough to bring tears to my eyes.
I hadn’t told Treblo which experiences were supposed to be important. I hadn’t chosen the lyrics or the arrangements. I hadn’t even named the songs.
The interpretations came from outside. My reaction came from inside. That’s what I wanted to find out.
Of course, there are limitations. Treblo didn’t literally read my website. It received a description prepared by ChatGPT, and that description inevitably reflected ChatGPT’s interpretation of the original material.
So this wasn’t a completely unfiltered machine reading of my entire online history.
It was one AI interpreting the writing and another AI interpreting that interpretation.
Which, when you think about it, is a fairly ridiculous way to make music. But it worked.
At least, it worked for me.
And now I’m curious what other people will hear in those songs. Maybe they’ll recognize something about themselves. Maybe they’ll hear something completely different. Maybe they’ll just enjoy the music.
I’m fine with any of those outcomes.
Now, there is going to be some commentary about real musicians losing opportunities to work. My push-back here is that I don’t know many musicians who know me well. And even if I did, having one read nearly 100,000 words and then interpret them into a song is purely fantasy.
This was a thought experiment—one that I was able to realize because of modern tools.
After all, the experiment was never about proving that the machines understood me, what other people might understand about me, or even what they might come to understand about me from those songs.
It was about finding out what they would say.
And this time, I’m not going any further down that rabbit hole. The songs exist, and I did find one that made me tear up. I won’t say I fully identify with all of them, but they did affect me. And that is enough.
For now.