Everyday dictation
Use the hold-to-talk workflow, choose a destination and understand when your words are inserted.
On this page
Hold, speak, release#
- Make sure TawkType is running. Its shortcut cannot work after you quit it.
- Click the field where you want the text: an email body, chat composer, document or editor.
- Check the active mode on the right of the status bar.
- Hold Right Ctrl, or your chosen push-to-talk key, and speak at a comfortable pace.
- Release the key and let go of any other modifiers. Keep the destination focused while recognition and cleanup finish.
- Review the inserted words before sending or publishing them.
The default minimum hold time is 250 milliseconds. A very short tap is ignored as accidental. Recognition speed varies by hardware, model and recording length; there is no fixed release-to-text time for every computer.
Get predictable results#
Start with short sentences to establish microphone placement and volume. Speak normally rather than spelling every word out. If a name is repeatedly misheard, save a spelling or explicit replacement instead of changing how you pronounce your entire sentence.
Punctuation is initially produced by the recognizer. Local cleanup fixes a limited set of fillers, spacing and capitalization; it does not reliably interpret spoken self-corrections or formatting commands. Optional Claude rewriting can interpret phrases such as “no, make that Tuesday,” but its output still deserves review.
Keep the destination in place#
TawkType checks the focused destination before delivery. If you switch windows while it is processing, it can withhold typing and copy the result instead. Some applications expose little useful information about their controls, so compatibility remains a best-effort process.
A window running as administrator may not accept synthetic input from a normal TawkType process. If the bar says the text was copied, focus the intended field and paste it yourself, or use the dictation box.
Stop or cancel#
Releasing the hold key finishes a recording and processes its text. With a configured toggle key, the second press finishes it. Pressing Esc cancels the active recording or pending processing rather than finishing it.
The recording limit defaults to five minutes. Reaching the limit finishes the recording so it can be transcribed; it is different from cancellation. Cancellation is not an undo command for text already delivered.
Read the status bar#
The bar shows your shortcut while resting, a level meter and elapsed time while recording, and processing states after release. Its controls open Settings, History and Copy last dictation; a recovery action appears when a completed result needs attention.
You can drag the bar to a convenient location. Its position is remembered, and it is designed not to take keyboard focus away from your destination. The app’s status text belongs on the bar, never inside your document.