// who built this

One person, talking more than he types.

I am Eric Disero. Audio engineer, then software developer. Solo and bootstrapped, building from the Comox Valley on Vancouver Island. I made TypeTalk because I needed it, and I use it dozens of times a day.

Eric Disero, founder of TypeTalk
Eric Disero
Founder · British Columbia, Canada

I started in audio. Years of making records, mixing sessions, living inside the small differences that nobody else hears. Then I moved into software, and these days I prompt Claude, Cursor, and ChatGPT for most of the working day. Which means I talk far more than I type. The whole job is describing what I want and watching a model build it.

Windows already ships dictation. Press Win+H and it listens. The problem is that it could not keep up with me. It needs the internet, so it dies the second my connection hiccups. It cannot place a comma to save its life. And the lag between pressing the key and seeing words appear was enough to break the flow I was trying to stay in. For a tool you reach for a hundred times a day, that gap is the whole experience.

So I tried the paid cloud apps. They were great in the trial. Fast, clean, exactly what the demo promised. Then I paid, kept using them, and watched them get slow and flaky once I was hooked. One of them sat there eating most of a gigabyte of memory while my fans spun up, just to wait for me to press a button. I read the reviews afterward and saw the same line over and over: quick during the trial, sluggish after you pay.

So I built the thing I actually wanted. Press a key, talk, and clean text appears instantly in whatever app has focus. Cursor, Claude, a terminal, my inbox, it does not care. The whole thing runs on my machine, so there is no server to wake up, no network leg to drop, and nothing to upload. It is fast on a plane. It is fast in a dead zone. It never quits on me. I am dictating this paragraph with it right now.

The two decisions

Why local, and why $99 once.

Local is the reason it is good

I did not build it on your machine to make a point about privacy. I built it that way because local is what makes it instant. No cloud handshake to wait for, no server that slows down when someone else gets busy, no connection to drop mid-sentence. It sits idle at under 200 MB of memory, so you forget it is running. Your voice staying on your laptop is just what falls out of building it right. Ctrl + Shift + Space, talk, done.

$99 once is the only honest price

Cloud tools charge monthly because every transcription costs them money. Theirs runs on a server they pay for, so the meter never stops. Mine runs on your hardware, which means it costs me nothing per use. There is no ongoing cost to cover, so charging you forever would just be rent. Pay once, own it, get every future update. That is not a marketing promise. It is the only price the architecture can support.

No board, no investors

You reach me directly.

There is no venture money behind this and no team to route you through. I build it, I support it, and I use it myself every single day. If you request a feature, it lands on my desk. If you hit a bug, I am the one who fixes it. That is the upside of one person who needs the tool to work as badly as you do.

Eric Disero, founder

Try the thing I built for myself.

Free to start, 2,000 words a week, works offline forever. Pay $99 once when you outgrow it.