Windows
- Ctrl Ctrl Double-tap Ctrl in any app. The Akoway pill pops up.
- Click the mic and talk. Click again when you're done.
- Your words land right where your cursor is.
Akoway turns your voice into clean text in any app on Windows and Android. It runs entirely on your device, and it removes your “um”s as you go.
Windows 10 / 11 · Android 8+ · No account, no subscription
Nobody speaks in perfect sentences. Akoway drops the “um”s, false starts and repeated words while it listens, so what lands in your text box is what you meant.
Imaginebusinessesthatmeasuresuccessumnotonlybyhowmuchtheyearn,butbyumhowmuchvaluetheytheycreate.
No ums. No repeats. Just what you meant.
Both are Underdog's open Bark 3.0 models, converted and tuned to run fast on your own device. Akoway downloads one once, on first launch. Switch any time in Settings on Windows.
Writes what you meant. Drops “um”, “uh”, false starts and repeats as you speak.
Writes exactly what you said, word for word, with punctuation and capitals.
Yes. No subscription, no trial, no premium tier. Akoway is MIT-licensed open source, and the speech models are open too.
Only once, to download the speech model the first time you open it. After that it works fully offline, even in airplane mode.
Never. Your audio is turned into text on your own device and then thrown away. The apps have no account and no analytics. (This website counts visits anonymously, without cookies, so we know what people find useful.)
Double-tap Ctrl from any app and the Akoway pill appears; click the mic and talk. You can also use Ctrl + Alt + F9, or set your own shortcut in Settings. Akoway just needs to be running: it sits quietly in the system tray, and the installer can start it with Windows so it's always ready.
Download the APK on your phone, open it, and allow installs from your browser when Android asks. Then open Akoway and follow the three setup steps. The optional floating tab uses Accessibility; if it's greyed out, open Settings › Apps › Akoway › ⋮ › Allow restricted settings.
English for now. Bark 3.0 is an English speech model.
Not yet. The code is open, so a Mac or Linux port is welcome: see it on GitHub.