The First Real Session

The app made sense until someone else touched it.

That is annoying to admit.

But it is true.

When I am building Ground Level Languages, I know what everything is supposed to mean.

I know why the Sound Garden exists.

I know why the Meaning Tree matters.

I know why I do not want to cover everything in text.

But that is the problem.

I know.

The user does not.

The First Real Session

I recently met someone in Taipei to record Mandarin audio for the app.

We rented a small room in Zhongshan for about an hour. Before the session, we were both early, so we waited downstairs in Family Mart and talked for a bit.

That helped.

It made the whole thing feel less awkward.

Still, I was nervous.

Not in a dramatic way.

More like:

I have never actually done this before.

I had written recording plans.

I had thought about what I wanted.

I had imagined how the audio would fit into the app.

But sitting in a room with someone and directing a recording is different.

Suddenly, the questions are not theoretical anymore.

Should I ask for another take?

Should this sound more natural?

Is this too much?

Is this boring?

Am I wasting her time?

That is a different kind of pressure.

The Voice Made It Real

The first recording session was for Mandarin.

That made sense.

I live in Taiwan.

I am learning Mandarin.

And the first demo is connected to the world I am actually living in.

The goal was not to record a whole course.

That would have been too much.

The goal was smaller:

Get enough audio to make the demo feel real.

Not complete.

Not perfect.

Real.

Once there was an actual voice, the app felt different.

Before that, the story was just text.

The vocabulary was just a list.

The lesson flow was just a plan.

But a voice changes things.

Suddenly there was something to build around.

Something to listen to.

Something that could actually live inside the app.

For a sound-first app, that matters.

The audio is not decoration.

It is the foundation.

Where the App Broke

After the recording, I also got to show the app.

That was useful.

And uncomfortable.

Some parts were confusing.

The navigation was not clear enough.

In my head, the flow had logic.

Start here.

Tap this.

Listen.

Move here.

Preview the language.

Enter the lesson.

But that logic was mostly living in my head.

The screen was not doing enough work.

The user should not need me standing next to them explaining what the app is supposed to do.

If I have to explain the interface out loud, then the interface is not finished.

What Actually Worked

The useful part is that not everything failed.

The language previews seemed interesting.

And some of the parts that felt like games were fun.

That stood out to me.

Because that is probably the direction.

Not more explanation by default.

Better interaction.

When something feels like a game, the user is more willing to try things.

They tap.

Something responds.

They notice what changed.

Then they try again.

That is basically the learning loop I want.

Not fake gamification.

Not points.

Not streaks.

Not turning language learning into a slot machine.

I mean the part of games where the system teaches you by responding to what you do.

That is what I want more of.

The Meaning Tree Problem

The Meaning Tree is where this matters most.

It is supposed to be the story side of the app.

The learner hears something before fully understanding it.

They see scenes.

They slowly connect sound to meaning.

The story gives the audio a reason to matter.

That is the idea.

But an idea is not the same as an experience.

If someone enters the Meaning Tree and is not sure what to do, then the idea is not working yet.

It does not matter how good the philosophy sounds.

The experience has to carry the learner.

Maybe the learner taps a fruit and a few story scenes appear.

Maybe they hear a line from the story before understanding it.

Maybe the screen shows just enough movement to make them want to continue.

The goal is not to explain everything.

The goal is to make the next step feel natural.

Tap.

Listen.

Notice.

Reveal.

Continue.

That is the kind of flow I want.

What I Learned

This session made two things clear.

First, the app needs real voices.

AI voices are useful, but I do not want the app to depend on perfect machine speech. Real voices have rhythm, breath, pauses, and small imperfections. They sound alive.

Second, the app needs clearer interaction.

Less text does not automatically make something better.

A quiet interface can still be confusing.

A pretty screen can still be useless.

A mysterious feature can still be dead decoration.

If the app is sound-first, then the design has to guide people toward listening without making them feel abandoned.

That balance is hard.

But it is also the point.

Where This Goes

The session made Ground Level Languages feel more real.

Not because everything suddenly worked.

It did not.

Not because the app was polished.

It was not.

But because the idea finally touched reality.

There was a real voice.

A real person using the app.

Real confusion.

Real feedback.

And that gave me something better than theory.

It gave me direction.

The next version needs to be clearer, more interactive, and more guided without becoming a wall of instructions.

Not because that sounded good in my head.

Because the first real session showed me what was missing.