HoldToType

Doku

Typing by voice when typing hurts

HoldToType was not written as an accessibility tool, and it does not control Windows by voice. What it does is narrower and useful: it turns speech into text at the caret in any program, with a single key or a toggle instead of a held chord, and it never sends your voice anywhere. Here is what that gives you, and what it does not.

Aktualisiert · Vitalii Yemets

Diese Seite ist noch nicht übersetzt. Die Startseite gibt es bereits auf Deutsch, der Rest folgt.

What it can do for you

  • One key instead of a chord. The shortcut is any combination you like, including a single key such as CapsLock or F9. Hold it, or switch to toggle mode: press once to start, once to stop, hands free in between.
  • Text in any program. Mail, chat, a browser field, a document: the words go where the caret is, as a paste or as keystrokes.
  • A plate you can read. While you speak, a small plate shows the state and, with Nemotron 3.5, the words as they arrive. It stays on top of other windows and never steals the focus.
  • Voice commands you define. Say new line, get a new line; say a phrase, get the text you attached to it. Punctuation comes from the model or from post-processing.
  • Your words. Names and terms in a dictionary, so you are not correcting the same word every day.
  • Nothing leaves the computer. No account, no cloud, no telemetry, which matters when the words are medical or private.

What it does not do

  • It does not move the mouse, click, open programs or read the screen aloud. For control by voice, Windows 11 has Voice Access, free and on-device in six languages; the two run side by side.
  • It is not tested with screen readers. The settings window is an ordinary web page with labels and a main landmark, but nobody has walked it with NVDA or Narrator yet. If you do, an issue on GitHub with what you found is welcome.
  • It does not correct by voice. You fix a word with the keyboard or the mouse, or you dictate the sentence again.

Setting it up to spare your hands

  1. Install from the download section.
  2. In Controls & behaviour switch the hotkey mode to toggle if holding is the problem; under Shortcuts set a single key you can reach.
  3. Pick Nemotron 3.5 in the wizard for live text on the plate, or Parakeet for speed; the models page explains the choice.
  4. Add the words that come up every day to the dictionary and define a few voice commands.
  5. Turn on post-processing so punctuation and fillers are handled for you.

Honest limits

Recognition happens on your processor; on an old laptop pick a small model and expect a second of delay after you let go. Long dictation works, but the program is built around a sentence or a paragraph at a time. The builds are not code-signed yet, so Windows warns once before the first launch. And it is one person's program: if something in it stands in your way, say so on GitHub, and it will most likely be fixed.

Questions

Can I use a single key?

Yes. Any combination or single key the program can capture, CapsLock and function keys included. Toggle mode removes holding altogether.

Does it work with Voice Access?

They do not conflict. Voice Access controls the PC; HoldToType dictates. Some people keep Voice Access for clicking and HoldToType for text.

Is it free for me?

Free for everyone, with no limits and no account, under the MIT licence.

Try it with one key

Install, set CapsLock or a function key under Shortcuts, and dictate a sentence.

Download HoldToType