Last time I wrote about deciding to start using AI. So here I was, ready to actually begin — and the very first thing I did was get ready to stop typing on a keyboard.
“Wait, this is supposed to be about AI, and you’re talking about input methods?” That is probably what you are thinking. But for me this turned out to be exactly the right call. Today I want to talk about the part of working with AI that people overlook the most: how you get your words in. If you are about to start with AI yourself, I think this one will be genuinely useful.
Talking to an AI means typing far more than you expect
Using AI comes down to one thing: giving instructions in writing. “Build this.” “Fix this part.” “Make it more like this.” Every bit of it goes through text. And if you want good results, your instructions tend to get specific, which means they get long.
I noticed it as soon as I started. You type. And then you keep typing.
I can type about as well as the average person. Even so, after a few hours of going back and forth with AI, my fingers got heavy. Worse, while I was busy typing out the thought in my head, I would forget the next thing I wanted to say. My typing speed simply could not keep up with my thinking speed. It sounds like a small thing. It was a real, constant source of stress.
The answer I found on YouTube: voice input
While I was looking for a way around this, I came across something on YouTube. The people who are good at using AI all use voice input.
When I thought about it, of course they do. People speak far faster than they type. If saying the thing out loud is enough to put it on the screen, then the whole “typing is exhausting” problem disappears. And if I can give instructions the way I talk, the conversation with the AI gets more natural too.
“This is it,” I thought. But there was one wall in the way.
Voice input has had a bad reputation for years. Full of misrecognized words, every “um” and “er” faithfully typed out, and fixing the result taking longer than typing it yourself in the first place. You have probably had that experience. I certainly had. So I was only half convinced.
Then I found Aqua Voice
I looked around, and the tool with the best reputation was a voice input program called Aqua Voice. It runs on a computer, both Windows and Mac.
It completely overturned my belief that voice input was useless. If I had to explain what makes it different in a way a beginner can follow:
It does not just turn sound into letters. It shapes it into what you were trying to say.
Ordinary voice input transcribes the noise you make, exactly as you made it. So the hesitations — the “um,” the “er” — land on the page along with everything else. Aqua Voice is different there. It strips out those fillers, adds punctuation on its own, and tidies the sentence as it writes it. What comes out is not what you said. It is close to the sentence you wanted to write. That difference is what made it comfortable.
It is also fast. It is not tied to one particular app — a browser search box, a chat input field, an email compose window, anywhere text can go. Which means my AI conversation window, and this blog, and everything else. Put the cursor where you want the words, hold a key, and talk.
Honestly, it moved me. The feeling of putting words down at the speed you think them. The sentences in my head flowing straight onto the screen. Most of this article was spoken into Aqua Voice.
Getting started was easy
Setup was not difficult. Download the software from the official site, install it, and register with a Google account. You can pick your language in the settings — it supports 49 languages, including Japanese — and once you give the program permission to use your computer’s microphone, you are ready to go.
Let me be straight about the price too. The free “Starter” plan lets you try it up to 1,000 words. After that you need the paid version, Pro, at around 8 dollars a month (roughly 1,200 yen). I tried it briefly, immediately felt it would pay for itself, and went paid. If you go back and forth with AI in volume every day, I think that investment comes back to you without question.
One honest caveat
I have written nothing but good things so far, so to be fair, here is the one warning.
Your audio is processed over the internet. So if you are about to speak your company’s confidential information, or personal details you would rather nobody knew, it is worth being a little careful. For what I use it for — the blog, chatting with AI, throwing ideas around — it is no problem at all.
If you are about to start with AI
If you are seriously thinking about using AI from here on, I would strongly suggest getting a voice input tool ready alongside the AI itself.
AI is a tool you operate by writing instructions. So the easier it becomes to produce those instructions, the easier the AI is to use. For me, Aqua Voice was the partner tool that belongs right next to the AI. Just having it makes working with AI dramatically more comfortable.
“Typing is a drag, so I drift away from the AI.” I think that is a place beginners stumble more often than you would guess. Which is exactly why you set your tools up first. The long way round is the short way.
Next time I will finally get to the first thing I actually built with AI: a record of my attempt to make a tool that automatically edits the pile of fishing videos I had let build up. I will be honest about what worked and about where I tripped up right at the start.
Note: the information in this article is from 2026. Software pricing and features change, so please check the official site for the latest details before you sign up.
If you want the background to this one, the story of how I got here is in the day I started with AI in my forties, and the project I mentioned above is in building a video editing tool with AI.
