Hands-free typing on a Mac with RSI

Written 15 August 2026.

If you are dictating because your hands hurt, a hold-to-talk hotkey is its own small problem. Here is what to set up instead.

This page is about mechanics rather than medicine. Nothing here is medical advice, and a tool that reduces keystrokes is not a treatment for an injury.

Stop holding the key

Hold-to-talk is a good default — it makes the microphone's state unambiguous and it is impossible to leave running by accident. It also means sustained pressure on one finger for the length of every sentence, which is the exact motion a lot of people are dictating to avoid.

The alternative is latching: a quick tap starts recording and leaves it on, another tap inserts the text, Escape discards it. In Bunso this is the hands-free setting, it is off by default, and holding still works normally when it is on. It is in the free tier, along with foot pedal support and the rest of the accessibility behaviour — none of it is behind the paid tier.

A foot pedal is the upgrade most people skip

A USB foot pedal presents itself as a keyboard, so it works with any tool that binds a hotkey — including this one. It moves the start/stop action off your hands entirely, and it keeps the thing hold-to-talk is good at: the microphone is on exactly while you are asking for it, with no state to forget.

If you have any wrist or finger strain and you dictate daily, this is a small amount of money for the single largest reduction in hand use available in the whole setup.

What happens if a latched recording is forgotten

Worth knowing, because it is the obvious risk of latching, and because a tool's answer tells you whether it was designed for this use or had it bolted on.

A latched recording has no key held down by definition, so the usual safety net — stop when the keys come back up — does not apply. What must keep applying is an elapsed-time cap, which is the only thing that ends a recording nobody came back to. Bunso keeps that cap active for latched recordings specifically for this reason.

Reducing the corrections, not just the typing

The keystrokes you save dictating are easily lost again fixing what came out. Three things move that number more than anything else:

• Teach it your vocabulary. Names, jargon and product names go in a custom dictionary, where they bias the recogniser before it transcribes. Why that works better than fixing the text afterwards.
• Let it learn your corrections. When you fix the same word repeatedly, that correction should become permanent without you managing a list.
• Save the paragraphs you retype. Anything you write more than twice — an address, a signature, a standard reply — can be stored under a spoken keyword and typed in full when you say it. That is the biggest single keystroke saving available, and it is not really dictation at all.

Where dictation still needs your hands

Being straight about the limits, because over-promising here wastes the time of people who have the least to spare:

• Editing is still manual. Selecting a word in the middle of a paragraph and changing it is a mouse-and-keyboard job in most tools. Bunso has voice editing — select text, speak an instruction, the text is rewritten in place — but it is in the paid tier and it does not remove the selection step.
• Setup needs your hands. Three macOS permission grants, and the settings.
• It does not drive your Mac. This is dictation, not voice control. macOS has Voice Control built in for navigating the interface by voice, it is free, and it is a genuinely different tool — the two are complementary rather than competing.