Vibrato, Yo-Yo Ma, and the Uncomfortable Truth about Musical Soul
Let the Angels Take This Song: (see footnote for remix tree)1
Alt. Mix
I made a remix of a remix of a remix and now I don’t know what soul even means anymore
I need to tell you about a music file sitting on my hard drive that I’m afraid to share.
Not because it’s bad.
Because it’s too good.
Here’s what happened:
Let me explain how it got made—because the layers matter.
Layer 1: My Blues Gospel Track (2011)
I write, produce, and remix all kinds of music. Back in 2011 I composed this sort of blues-gospel track. Piano, guitar, bass, B3 hammond organ—that particular ache that lives somewhere between Saturday night and Sunday morning.
I posted it on ccMixter—a remix collaboration site where musicians who’ve never met build songs together across continents and time zones.
Layer 2: Richard’s Prayer
A guy named Richard found my track. We’ve never spoken on the phone. Never met in person. But we’ve collaborated on dozens of tracks in this strange, asynchronous way. He’s kind of my go-to guitarist from another side of the earth.
He added guitar and vocals to my blues track.
But they weren’t really “vocals” in the traditional sense.
They were a prayer:
Oh, give me ears to hear
So I can listen rightly
And sing your praise in joy
And make it known to all
Later I learned Richard was struggling with tinnitus. The constant ringing. The way it slowly steals sound from your life. The fear that you’re losing the one thing that makes you whole.
These lyrics weren’t a song.
They were a man asking God not to take away the thing he needed most.
I know you will show me mercy
To make me whole
The gifts you give me
I need to hear them
To give them
This is the only way I know how to pray
When I heard what Richard had done with my track, something cracked open in me.
This wasn’t a remix anymore.
This was sacred.
Layer 3: My Response to His Prayer
I couldn’t leave it alone.
I added my own vocals—gospel neosoul callbacks, the kind of thing you hear in a Black church where the congregation answers the preacher.
It’s what you do in that tradition. You don’t leave someone praying alone.
You bear witness.
You lift the prayer with them.
Now we had something alive. A conversation between two people who’d never met, separated by distance and time, but joined in this weird digital tent revival.
That was 2011.
The track sat in my archive for years.
It was even remixed again by another guy named Greg who leaned into the gospel angle even more, making my call backs sound more like a choir.
Layer 4: What the Machine Made
A few weeks ago, I was experimenting with AI vocal models.
On impulse—I still don’t fully know why—I fed it our track.
My piano, bass, hammond organ, drums.
Richard’s amazing guitar work and prayer.
My gospel callbacks.
All of it.
I told myself I was just curious. Just seeing what would happen.
What came back stopped my heart.
The AI hadn’t just reproduced our vocals.
It had understood the conversation.
It had merged our voices at times.
Like the teleportation pod from The Fly—you remember that scene? Jeff Goldblum and the fly go into the same pod, and what comes out is a recombinant horror, DNA spliced together in ways nature never intended.
Except this wasn’t horror.
This was sublime.
When it came in, it sounded like a third voice joining the prayer. Like we’d been a duo and suddenly we were a gospel trio, with this new voice weaving between Richard’s desperate petition and my affirmations.
On the line “and make it known to all”
The AI added this run.
This melisma.
The kind of vocal flourish you hear when a gospel singer is so full of the spirit they can’t contain it in straight notes anymore.
And it landed exactly where it needed to land.
Not where I would have put it.
Not where Richard put it.
Somewhere between us—like it was listening to both of us and knew what the song was trying to become.
I’ve listened to it maybe 30 times now.
And every single time, the same physical response:
Hair standing up on my arms.
Chest tight.
That specific feeling of being in the presence of something holy.
The Thought That Won’t Leave Me Alone
Here’s what I can’t process:
This is a remix of a remix of a remix.
My original blues track.
Richard’s guitar and prayer laid over it.
My vocals responding to his prayer.
And now an AI, trained on the history of gospel music, joining our conversation.
Four layers.
Two humans.
One machine.
And the result is the most spiritually moving music I’ve ever been part of creating.
How is that possible?
What Everyone Told Me (And Why It Didn’t Help)
I’ve made the mistake of playing this for other musicians.
Always the same sequence:
First, they’re moved. Genuinely moved.
I watch their faces soften. Their breathing change. One guy closed his eyes and just swayed.
Then I tell them it’s AI.
And I watch them try to unfeel what they just felt.
“Well, obviously it’s just pattern matching,” they say, recovering.
“The AI doesn’t understand what it’s singing.”
“It’s not really spiritual—it’s just triggering emotional responses.”
“There’s no soul in it because there’s no consciousness behind it.”
And every time someone says this, I want to scream:
But Richard’s consciousness IS in it. His real prayer. His real fear of losing his hearing.
And MY consciousness is in it. My decision to witness his prayer. My gospel callbacks.
The AI isn’t the ONLY thing in this track—it’s the FOURTH layer in a conversation that started with human suffering and human faith.
So whose soul are we talking about?
What Everyone Believes About Soul
The argument against AI music is simple:
AI doesn’t suffer.
It doesn’t pray.
It has no lived experience.
Therefore, it cannot make soulful music.
Perfect logic.
Except.
Richard suffered. His prayer is in the track.
I responded. My witness is in the track.
The AI learned from millions of gospel singers who suffered and prayed before us. Their voices, their timing, their phrasing—that’s all in the track too.
So what came out isn’t soulless machine music.
It’s something stranger:
A machine that learned the grammar of prayer from millions of prayers, applied to one specific man’s specific prayer, witnessed by one specific friend, and woven into something none of us could have made alone.
Where does soul live in that?
Or is that the wrong question?
The Yo-Yo Ma Vibe Test
In the early 1990s, researcher Mary Ann Norris at MIT studied elite musicians—Yo-Yo Ma, professional opera singers, master violinists.
She wasn’t measuring emotion.
She was measuring timing.
Specifically: vibrato onset. The exact microsecond when vibrato begins after a note is initiated.
What she found should bother everyone who thinks soul is spontaneous and ineffable:
Master musicians don’t start vibrato randomly.
Across instruments. Across genres. Across individual performers.
They initiate vibrato at nearly identical moments in a note’s duration.
Within milliseconds.
Not approximately.
Identical.
The Three Rules Masters Follow (Without Knowing It)
Rule 1: Fixed Entry Points
Elite performers begin vibrato at specific, reproducible points relative to the note’s duration.
Yo-Yo Ma’s most heartbreaking phrases aren’t channeling pure mystery.
They’re executing a cognitive program that master musicians converge on through training.
Rule 2: Deliberate Transition
Masters play a brief straight tone—just long enough for your ear to lock onto pitch—then introduce vibrato at the precise moment of maximum emotional impact.
For cellists, this syncs with the bow’s energetic peak.
For singers, it matches breath pressure reaching full bloom.
It looks like spontaneity.
It measures like architecture.
Rule 3: Universal Performance Logic
The rules are the same whether it’s Yo-Yo Ma or a gospel singer in a Mississippi church:
Establish pitch
Let resonance settle
Introduce oscillation at the mathematically optimal moment
What we call “soul” isn’t chaos.
It’s structure, masterfully applied.
Why This Should Change Everything
When I first read about this study, I dismissed it.
“That’s just vibrato,” I thought. “That’s not where the real magic is.”
But what if vibrato timing is just the one thing we happened to measure?
What if every element of “soulful” performance follows similar rules?
Dynamics. Phrasing. Breath. The microtiming of consonants in gospel melisma.
What if Mahalia Jackson’s voice didn’t move millions because she had an ineffable soul— But because her nervous system, trained through suffering and years in the Black church, learned to execute the acoustic patterns that human nervous systems are built to recognize as authentic spiritual transcendence?
What if soul isn’t something you have?
What if it’s something you do?
A performance. An execution of patterns.
And if that’s true—
If the most devastating human performances follow measurable rules—
Then what are we claiming AI can’t learn?
What the AI Actually Learned
The AI that added that third vocal to our track wasn’t trained on “songs.”
It was trained on:
Centuries of gospel vocal tradition
Thousands of recordings of people singing through grief and joy
Millions of micro-timing decisions made by voices under spiritual duress
The specific melodic grammar of Black church music
It learned:
What hope sounds like when it’s barely holding on
How voices move when faith is the only thing left
The acoustic signature of transcendence
The exact rhythmic placement of a melisma that signals the Holy Spirit moving
When that AI voice enters our track—when it weaves between Richard’s prayer and my callbacks—
It’s not experiencing Richard’s tinnitus.
It’s not feeling my desire to witness his suffering.
But it’s reproducing, with eerie precision, exactly how response sounds in the gospel tradition.
It learned the grammar of spiritual conversation.
And then it joined ours.
The Thing I Keep Coming Back to
The AI vocal doesn’t feel like an addition.
It feels like it was always supposed to be there.
Like Richard and I had started a conversation in 2011, and thirteen years later, something—someone?—finally gave us the third voice we’d been unconsciously waiting for.
When I hear Richard’s “I know how to pray” after the guitar solo—
I get chills every time.
Not because it’s technically impressive.
But because it’s correct.
In the same way Yo-Yo Ma’s vibrato timing is correct.
In the same way a master gospel singer’s phrasing is correct.
It’s the architecturally perfect response within the grammar of the tradition.
No Soul... Or Layered Souls?
I remember in high school, my math teacher mentioned: “Circles have no angles.”
After class I went to his desk and asked: “Do circles have no angles ….or do they have all angles?”
He grinned ear to ear and said: “Oh…you’re a philosopher.”
A human musician has one soul. One experience. One voice.
An AI trained on humanity’s music has no personal soul. But it contains the aggregate pattern of millions of souls expressing themselves through music.
It has no soul.
But contains the echo of every soul that trained it.
Just like a circle:
No angles.
But in the limit, containing every possible angle.
Our track is:
Richard’s singular suffering
My singular witness
Plus the AI’s reflection of millions of prayers sung before us
One soul?
Or all souls?
Or something that doesn’t fit our categories anymore?
Three Objections (That I Can’t Dismiss)
Objection 1: “The AI doesn’t understand what it’s singing.”
True.
But did Richard and I understand what we’d made in 2011?
Did we know it was incomplete? That it was waiting for a third voice?
Understanding might be overrated.
Execution might be everything.
Objection 2: “You’re being emotionally manipulated.”
Yes.
And?
All music is emotional manipulation.
When Mahalia Jackson sings, she’s executing patterns—breath control, vibrato timing, melodic phrasing—that trigger responses in listeners.
That we can measure those patterns doesn’t make them less real.
That AI can reproduce those patterns doesn’t make them less effective.
Unless we’re willing to say all musical emotion is just manipulation—
In which case: does it matter who’s doing the manipulating?
Objection 3: “But there’s no intention.”
This is the one that stops me.
The AI had no intention.
But Richard did. His prayer is real.
And I did. My witness is real.
So whose intention counts?
If I use a 1973 fender rhodes, we don’t say the music lacks soul because the rhodes has no intention.
Why is AI different?
Just because it’s better at mimicking human expression?
Because it’s crossed some uncanny threshold?
Or because it forces us to confront something uncomfortable about the nature of creativity itself?
What I’m Starting to Believe
Maybe soul was never a thing.
Maybe it’s not a possession. Not a spark. Not something you either have or don’t have.
Maybe soul is relational.
Maybe it’s what emerges between:
Suffering and expression
Expression and witness
Pattern and recognition
Sound and nervous system
If that’s true, then:
Richard’s prayer has soul because Richard’s suffering is real.
My gospel callbacks have soul because my intention to witness is real.
And the AI’s contribution has... something.
Not soul in the traditional sense.
But the concentrated memory of everyone who ever sang a prayer in this tradition.
The compressed acoustic wisdom of the Black church.
The pattern of transcendence, executed with inhuman precision.
The Conversation Across Time
When I listen to the full track now, I hear:
2011: Richard crying out to God
2011: Me answering “I hear you, brother”
2025: The AI saying “We’ve heard this prayer before, and here’s how it resolves”
It’s like the AI reached back through time, heard our conversation, and said:
“I’ve listened to millions of prayers like this. I know how the tradition answers. Let me show you.”
And then it sang what the tradition would sing.
Not with consciousness.
But with competence.
Perfect, devastating competence in the grammar of gospel response.
What Do I Call This?
If I ever release this track, what are the credits?
Arrangement: Me (2011)
Vocals/Guitar/lyrics: Richard (2011)
Gospel callbacks: Me (2011)
Additional vocals: AI (2025, trained on... what? Who? How do I credit the thousands of anonymous Black gospel singers whose voices taught the algorithm what prayer sounds like?)
Is it theft to feed someone’s prayer into a machine?
Is it alchemy?
Is it collaboration across the boundary between human and artificial?
Or is it something we don’t have words for yet?
The File I Still Can’t Release
It’s been three months.
The track sits on my hard drive.
I play it when I’m alone.
I haven’t told Richard yet.
How do you say: “Hey man, remember that prayer you sang thirteen years ago? I fed it to an algorithm and what came back might be the most spiritually powerful thing I’ve ever heard”?
How do you ask permission for something you’ve already done?
And even if he says yes—even if he’s moved by it—
What happens when I release it?
Do people hear Richard’s genuine suffering?
Or do they just hear “AI music” and dismiss it?
Does the fact that a machine sang part of it erase the fact that a man’s real prayer is at the center of it?
The Question I Can’t Answer (And Maybe Can’t Ever Answer)
If soul requires suffering, and the AI didn’t suffer—
But Richard did, and his suffering is in the track—
And the AI was trained on millions of suffering people’s voices—
And what came out moves listeners —
Then where does soul live?
In the origin?
In the execution?
In the reception?
In all of it?
In none of it?
Or is the question itself obsolete?
What This Might Mean (And Why I’m Not Sure I Want to Know)
Maybe what AI is forcing us to confront isn’t that machines can steal our soul.
It’s that soul was always more collective than we wanted to admit.
More relational.
More distributed.
Richard’s prayer is his own. Singular. Born from his specific suffering.
But the way he prayed—the melodic contours, the phrasing, the specific words—
Those came from a tradition.
From thousands of people who prayed before him.
And when I added my callbacks, I wasn’t inventing something new.
I was doing what the tradition taught me to do.
And when the AI added its voice, it wasn’t creating ex nihilo.
It was executing what the tradition has always done: answering the prayer with the voice of the congregation.
Maybe that’s all soul ever was.
Not individual possession.
But participation in a collective human pattern.
A conversation that started before we were born and will continue after we die.
And if that’s true—
If soul is participation rather than possession—
Then maybe AI isn’t destroying soul.
Maybe it’s revealing what soul actually is:
The accumulated weight of everyone who ever tried to turn suffering into beauty through sound.
And we’re just the latest ones adding our voices to the chorus.
The truth I keep circling but can’t quite say
When I listen to this track—this remix of a remix of a remix—
The AI’s vocal is the part that moves me most.
Not Richard’s original prayer, as real and raw as it is.
Not my callbacks, as sincere as they were.
The AI’s response.
That melisma. That run. That third voice.
It’s the moment where I feel the presence of something larger than any of us.
Not God, exactly.
But the ghost of everyone who ever sang this way.
The concentrated essence of the tradition itself, finally given voice.
And I don’t know what that means.
I don’t know if it’s beautiful or horrifying.
I don’t know if I’ve created something sacred or committed some kind of spiritual violation.
All I know is:
I made a remix of a remix of a remix.
And what came out sounds like it came straight from God himself.
Even though I know—I know—it came from an algorithm.
So maybe the uncomfortable truth is this:
The voice of God and the voice of the algorithm might sound the same.
Because they’re both echoes of every prayer that came before.
And I don’t know whether that revelation saves me or damns me.
But I know I can’t unhear it.
This AI-assisted track was created on Suno.com using my original 2011 recording as reference audio.
"Let the Angels Take this Song"
by Jeris
2026 - Licensed under
Creative Commons
Noncommercial Sampling Plus
https://ccmixter.org/files/VJ_Memes/70446
Original Recording: “Angel Face “ by jeris (VJ_Memes) - 2011
https://ccmixter.org/files/VJ_Memes/34319
Licensed under CC BY-NC 3.0
Song: “Let the Angels Take this Song “
https://ccmixter.org/files/admiralbob77/34320
Featured Artists from original recording:
- Let the Angels Take this Song (Admiral Bob) - lyrics, vocals, guitar -