My 10-Year Journey with Voice Input, After My Subordinates Told Me They Could 'Decipher' My Messages

For over 10 years, probably since around 2013, I was a pretty suspicious character, muttering to myself in public.

Back then, smartphone voice input was still a parade of typos and omissions.

I want to write about how I finally reached the pinnacle of 'brain-dumping' after encountering the latest AI voice tools.

Those days in Ebisu when I was being 'deciphered'

This is a story from back when I was still an employee at a startup, working energetically like a workhorse (?) before I went independent.

My office at the time was inside Yebisu Garden Place.

It was past that moving walkway, the 'Sky Walk,' which is quite a walk from Ebisu Station.

I was so pressed for time even while moving that I would constantly dictate work notes and messages into my iPhone while walking.

Things like, 'Regarding such-and-such, I want to do this by such-and-such date, so please convey this to so-and-so, and once you get confirmation, please send a formal request to so-and-so.'

Just like that. Nothing special, just common business communications and requests.

But the accuracy of voice input back then, looking back, was truly, endearingly useless.

'Regarding such-and-such, please check the uterus.'

Uterus.

No, no, what kind of emergency is that?

Every time I tried to convey something while on the move, I would send those kinds of typos—which would be an instant 'out' today—directly to my colleagues and subordinates without even checking them.

I didn't send them to clients or important business partners, of course. (I'd make excuses, saying, 'It's only for internal chat messages, at least.') But my colleagues must have been receiving some pretty suspicious messages on a daily basis.

Even so, the people around me at the time were truly kind and warm-hearted.

'Reiko-san, it's fine. We understand what you're trying to say.'
'We've managed to decipher this, so please don't worry about it!'

Decipher.

Before I knew it, my work instructions and requests had become something like ancient Egyptian hieroglyphs.

Under normal circumstances, I should be sending proper, error-free text. That is a given, and it is not as if I ever thought, 'Oh well, this is fine.' However, with the technology at the time, I had no time to check, and a certain amount of typos was unavoidable as the price for sending messages while on the move. I always felt sorry that others were kindly absorbing the consequences of that.

That is precisely why I am where I am today.

With the addition of AI to voice input, my 'messages that require deciphering' have dropped to almost zero.

I no longer have to rely on someone else's kindness. I no longer have to cause anyone extra trouble.

That alone has made me feel so much more at ease.

I believe this is what it means for technology to evolve.

And yet, here we are.

Now, I can just mutter into my smartphone, haphazardly and with disjointed subjects and predicates, 'Ah, about that matter, actually let's go with that, oh wait, let's do this instead.'

The AI will then ask, 'You mean this, right?' and translate it into a perfectly polished, beautiful Japanese business email.

Is this really happening on the same planet?

I am simply trembling at the evolution of technology.

To be fair, the accuracy of the iPhone microphone itself has certainly improved with every upgrade (though I am not sure if that is the direct reason). The frequency with which 'deciphering' was required decreased year by year.

But the moment AI was added to the mix, the story changed.

It is not just a matter of accuracy improving. It feels as if the very stage of convenience has been raised significantly.

The Three-Way Battle: AquaVoice, SuperWhisper, and Typeless

As someone with a long history of using voice input, I have recently been using a few different tools depending on the situation.

First, when speed is the priority, I use 'AquaVoice'.

While holding down the designated shortcut key, the text is sucked into the screen one after another as I speak.

The moment I release the button, it instantly becomes text.

I find this sense of speed, where the system responds as soon as I speak, to be great for keeping my brain moving when I am working hard at my PC.

On the other hand, I also used a tool called SuperWhisper (though I don't use it anymore).

It's quite convenient for outputting short sentences in bullet points or setting individual modes according to the purpose. You can pre-configure the output patterns you want, and when you dictate, you select which format you want to use, and it outputs in that style. Whether you want bullet points, a message-like format, or a specific template, it's (or was) convenient.

However, the timing of installing this tool was bad.
I tried to introduce it during my busiest period at work.

The setup video was very easy to understand, but I didn't have the energy left at the time to watch it and set it up properly.
In the end, I just kept using it in a half-baked way.

While dragging the feeling of 'I don't feel like I'm using this properly,' I continued to go back and forth, mainly using the convenient AquaVoice while occasionally pulling out SuperWhisper.

What I learned from this is the all-too-obvious lesson that 'when introducing a new tool, the enthusiasm and design at the start are important.'

And now, another one has been added: a tool called 'Typeless'.

There is one reason why I introduced this immediately.

Because it supported smartphones.

Both AquaVoice and SuperWhisper were mainly for PC use. But the moment I most want to use voice input is on my smartphone while on the move, specifically that moment while walking through the Ebisu Skywalk.

When you install Typeless as a custom keyboard on your iPhone, you can use voice input in the input fields of any app.

I thought, 'This is it.'

Even after subtracting the slight hassle (which is quietly annoying) of unlocking the smartphone, selecting the app in the input field, and waiting a split second to start dictating... the convenience is more than worth it.

It takes the incoherent content I speak and neatly summarizes and organizes it for output.

For now (as of the end of June 2026), I am using both Aqua Voice and Typeless depending on the situation.

▼ The tools are here: Aqua Voice/ SuperWhisper/Typeless

Stop 'studying' just to hoard information, and circulate your thoughts back into Obsidian

I first throw the thoughts from my brain (sometimes including trash) and raw gems that I've outputted with Typeless into an app called 'Drafts'.

Then, I link only the emotions, diary entries, and ideas that I think 'I want to keep properly' from among them to a note-taking app called Obsidian and stock them there.

Actually, I have always had a very bad habit of feeling satisfied just by buying reference books or books, as if I had already studied them.

The same goes for notes.

I would just write down whatever came to mind, leave it in a folder, and never look at it again.

I was drowning in a sea of self-satisfaction, thinking, 'Wow, I'm amazing for taking so many notes!'

But in the age of AI, I realized something.

In order for AI to think on my behalf, it is meaningless unless the raw data of my own thoughts is neatly stocked in one place first.

That is why I decided to start organizing Obsidian as 'the cleanest drawer of my brain'.

I pour out all the aimless, fuzzy thoughts in my head at once using voice input.

Then, I have the AI tidy them up just a little bit and gently place them in Obsidian.

This cycle makes my often-cluttered brain surprisingly clear.

My 'Voice' Infrastructure, Defeated by Air Conditioning

Just as I was enjoying my voice-input life like that, a sudden tragedy struck.

It was one day after summer began and we started blasting the air conditioning in the office.

My throat hurt.

Even swallowing saliva was a bit painful.

Usually, when I found typing on a keyboard too troublesome, I could just say, 'Hey, do this for me,' to my smartphone and be done with it.

But I had reached a state where just producing a voice was a bit difficult.

At that moment, I was somehow in trouble.

I realized that the premise that 'it's easier to use my voice than to type with my hands' was only barely holding up thanks to the health of my throat.

No matter how excellent the AI tool is, if my 'voice' infrastructure is cut off, it just becomes a paperweight.

Since then, I have been desperately taking care of my throat.

I keep throat spray on hand, drink water frequently, and suck on throat lozenges of suspicious colors.

Who would have imagined that the first step to mastering AI would be 'throat hydration'?

But I feel like this is actually something very essential.

When AirPods first came out, we saw people walking around town with white 'noodles' hanging from their ears, talking to themselves, and we were a bit put off, thinking, 'Wait, what's up with that person...?'

But now, that sight has become commonplace.

Voice input is the same.

The era where everyone naturally mutters their thoughts into their smartphones is just around the corner (or maybe it's already here).

That is precisely why we must take better care of our 'voice'—our most primitive and important tool for output.

No matter how smart AI becomes, I must continue to protect the 'voice' I use to pour out my thoughts and the 'place' where I store them.

The more technology evolves, the more it seems we return in the end to our own bodies, the most analog existence of all.

Today, too, I give my throat a quick spray and whisper to my smartphone.

'Hey, help me organize what's in my head today.'

This has turned into a rambling piece of writing, but since convenient tools are constantly emerging, I hope to continue using them well to give shape to my 'raw thoughts' in the future. Also, take care of your throat.