Ears First, Theory Later — Why an Ear Training App Is Needed
Even When AI Nails the Sound
The MUEDear development story is usually told as "it was supposed to be a quick build" — the in-house DSP implementation saga. I'd like to approach it from a different angle. Why ear training in the first place?
You know how AI mastering is becoming mainstream, right? We're doing AI mixing with MUEDial too. And the output accuracy of AI just keeps getting better.
Right.
But here's the thing — even if AI produces a perfectly accurate sound, who decides whether it's "good" or "bad"?
...The human.
“Even if AI delivers a perfect 100-point mix, if you're listening with 50-point ears, you can't tell the difference.”
— kimny
Exactly. Even if AI delivers a perfect 100-point mix, if you're listening with 50-point ears, you can't tell the difference. You just go "yeah, I guess it sounds better?" and that's it. You're not actually leveraging what the AI can do.
Even as the tools get more precise, it's meaningless if the user's perception can't keep up.
Critical listening is a human job. Sonarworks said the same thing in their 2026 study. No matter how far AI evolves, the ears that make the final "this is good to go" call are human.
Knowing Theory Doesn't Mean You Can Hear It
Can't you solve this by studying theory? EQ frequency bands, compressor behavior, that sort of thing.
Knowledge and hearing are two completely different things.
How so?
You might know that "boosting 2kHz increases vocal presence." That's textbook stuff. But can you actually listen to two versions — one with a 3dB boost at 2kHz and one without — and go, "yep, that's a 2kHz boost"? That's a completely different skill.
The gap between knowing and hearing.
That was me, you know. I learned theory from books and the internet. But I didn't actually start "hearing" things until after years and years of listening in studios.
It took time.
A ridiculous amount of time. And there was no efficient method for it. Just listen. Compare. Listen again. There was no structured training approach — you just had to grind it out on the job.
Your Ears Are a Physical Skill
And that's what the app aims to streamline.
But here's the key thing — you can't just ask AI to "train my ears" and have it work.
Why not?
Because ear training is a physical skill. It's like working out. AI can tell you "here's the correct push-up form," but you're the one who has to actually do the push-ups. The ability to distinguish sounds only develops through repeated listening with your own ears.
A domain AI can't replace.
When I spent two or three years stripping down the LMS features, ear training was the last thing standing. Auth, servers — all of that gets absorbed by AI. But ear training survived. Because anything that requires your body, you have to do yourself.
The "what's left after you strip everything away" idea — you talked about that in "Why a Composer Is Building Apps" too.
I keep arriving at the same conclusion. Only the irreplaceable things survive.
How I Trained My Own Ears
How did you train your ears, kimny?
On actual recording sessions.
Specifically?
During recording and mix checks, being able to catch "wait, did something change slightly here?" — that directly builds trust. If you can't notice, people start wondering "can I really trust this person?" So you listen. Intensely.
Pressure-driven.
More like professional survival instinct. When a mix revision comes back, can you instantly tell "ah, they changed this part"? If you can't, you can't do the job.
And you're trying to recreate that experience in an app.
“The foundational listening skills that pros develop over years in the field — that's what the app compresses into repeatable training.”
— kimny
Not all of it. You can't replicate the tension of a real session in an app. But the core exercise — listen to A and B, spot the difference — you can repeat that hundreds of times. In a real session, you might get a few comparisons per song. In the app, you can do ten in a minute.
Increasing the density of repetition.
Right. The foundational listening skills that pros develop over years in the field — that's what the app compresses into repeatable training. The top-level stuff still requires real sessions, but you can build the foundation with an app.
The Better AI Gets, the More You Need Your Ears
Here's the paradox — does demand for ear training actually increase as AI improves?
I think it does.
Why?
As AI mastering and AI mixing become the norm, AI handles the actual processing. But the person who decides "this is good enough" on the AI's output — that's still human, right? Without that judgment, no matter what AI produces, you just end up going "sounds okay I guess?" and calling it done.
The more the tools evolve, the more your ears are put to the test.
It was the same with plugins. You could own 50 different EQs, but if you don't have the ears to judge "which one fits this track," they're useless. Same with AI. In fact, as AI output quality goes up, the ability to hear subtle differences and make calls — that becomes the real differentiator.
And that's MUEDear's territory.
Exactly. MUEDear is a tool for creators to build their own judgment. Whether you let AI handle things or do it yourself is a separate question — either way, "being able to tell good from bad with your own ears" is the prerequisite. MUEDear builds that prerequisite.
Connecting to "The Rumination of Memory"
This connects to the "rumination of memory" idea from "The Brain Is a Playback Device," doesn't it?
Hearing ability is built on accumulated experience. You engrave thousands of patterns into your brain — "this sound sounds like this" — and when you hear something new, you match it against those patterns.
Searching and matching against a database of memories.
That's why beginners go "something feels different, but I can't tell what." Their database is too small — there's nothing to match against. MUEDear is a tool for efficiently building that database.
Practice (Melete) leads to memory (Mneme) leads to creation (Aoide). The same arc as the original Muses you discussed in the MUED article.
Exactly. MUEDear embodies Melete — the Muse of practice. Listen over and over, engrave the patterns into your brain. Not mysticism. Methodology.
This article was reconstructed from multiple conversations with AI (Claude).