There is a specific moment people describe, and it is rarely dramatic. Not an injury — a realisation. You are twenty minutes into a long reply and your forearm is aching in a way it did not use to, and you notice you have been holding your wrist slightly off the desk to avoid the part that hurts.
For a lot of knowledge workers, that is the week they start looking up dictation software.
To be clear about what this is: practical notes on reducing keystrokes, written by someone who builds a dictation tool — not medical advice. Repetitive strain injuries have causes that software cannot touch: posture, desk height, hours, grip, stress, and conditions that need a diagnosis. If your hands or wrists hurt, see a clinician. Read this as one lever, not a treatment.
Why the keyboard is such a stubborn problem
Typing is the one activity a knowledge worker cannot easily reduce, because it is not one task — it is the interface to all of them. Cut the emails and there is Slack. Cut Slack and there are tickets, commit messages, documents, comments, forms. The volume hides in the small stuff, which is exactly why “type less” is such useless advice.
The genuinely useful question is narrower: which keystrokes are load-bearing, and which are just transcription?
Writing code is load-bearing. So is precise editing — moving a clause, fixing one character, driving shortcuts. But a 200-word Slack explanation is not editing; it is you converting a fully-formed thought into keystrokes. That conversion is the part a microphone can do instead, and for most people it is the bulk of the daily total.
A hands-light setup on macOS
Nobody sensibly goes from all-keyboard to all-voice in a day. What works is layering, and macOS has more of the pieces built in than people realise.
Layer 1: Voice Control for navigation
Apple's Voice Control (System Settings → Accessibility → Voice Control) operates the machine by voice: clicking, scrolling, switching apps, working menus. It is a genuine accessibility feature, not a novelty, and it is the right tool for the part of your day that is pointing and clicking rather than writing. Worth knowing it exists even if you only enable it on bad days.
Layer 2: a dictation tool for the writing
This is where the volume is. The important property is not accuracy — it is how much editing the output needs, because editing is keystrokes, and keystrokes are the thing you are trying to spend fewer of. A dictation tool that hands you a transcript full of “um”s to clean up has moved your work, not reduced it.
That is the specific reason cleanup matters here more than anywhere else. Air Wisper transcribes on-device, rewrites the transcript — filler words removed, punctuation placed, self-corrections resolved — and types the finished text into whatever app you are in. The ratio that matters is words spoken to keys pressed afterwards, and it should be close to zero.
Layer 3: keep your hands for what they are good at
Precision. Selecting a word, hitting a shortcut, nudging a cursor. Trying to do fine editing by voice is slower, more frustrating, and the fastest route to abandoning the whole approach. Draft by voice, adjust by hand.
Layer 4: the physical layer
Software is one input among several. A better keyboard, a different mouse or trackball, monitor height, chair, and actual breaks all matter, and a physiotherapist or occupational health assessment will identify things no app can. Dictation reduces volume; it does not fix a workstation.
What the first two weeks are actually like
Expectations help, because the adjustment is real and short.
Days 1–2 feel wrong. Speaking a sentence you would normally type is self-conscious, and you will stop mid-thought to check the screen. Almost everyone's first-day verdict is “this is slower”. It is — for two days.
Days 3–5, the pause disappears. You stop composing before speaking and start speaking to compose. This is the actual skill, and it arrives suddenly.
Week 2, it becomes invisible. You reach for the shortcut without deciding to, the way you reach for copy and paste. This is also when the load change becomes noticeable — not as less pain necessarily, but as a day that ends with less accumulated strain.
Someone documented this arc in detail in I replaced typing with voice for a week.
Practical things that make it work
- Get a microphone off the desk. AirPods or any headset with a boom mic. Better input means fewer recognition errors, and fewer errors means fewer corrective keystrokes — which is the entire point.
- Start with the highest-volume, lowest-stakes text. Slack and internal notes. Do not debut voice on something that must be perfect.
- Do not stop when you misspeak. Say “actually, make that Thursday” and continue. Backtracking with the keyboard defeats the purpose.
- Treat your voice like your hands. It fatigues. Water, and pauses between long dictations.
- Solve the office problem honestly. Open-plan dictation is genuinely awkward. A booth, a quiet corner, or reserving voice for home days is a reasonable answer — partial adoption still reduces volume.
- Ask for it formally. Input tooling is a routine workplace adjustment, and asking through occupational health rather than IT tends to go better.
Where dictation genuinely does not help
Being straight about the limits is more useful than overselling:
- Code syntax. Brackets, operators, and camelCase are faster typed. Dictate the comments and commit messages instead — see 6 ways developers use voice to text while coding.
- Spreadsheets and precise navigation. Cell-by-cell work is a pointing task, not a writing one.
- Anything you cannot say out loud. Shared offices, confidential conversations, and passwords are keyboard territory.
- The underlying cause. If pain persists or worsens, software is not the answer you need. See someone.
The honest framing: dictation will not fix RSI. What it can do is remove a large share of the keystrokes that are pure transcription — and for people whose hands are already at their limit, changing the volume is worth more than any single ergonomic upgrade.
Frequently asked questions
Can voice to text help with RSI or wrist pain?
It can reduce how many keystrokes your day requires, which is the part of the load dictation is able to change. It is not a treatment and it does not address posture, workstation setup, or the underlying cause — those need a clinician. Think of it as one lever among several.
What is the difference between Voice Control and dictation on a Mac?
Voice Control is Apple's accessibility feature for operating the whole computer by voice — clicking, scrolling, switching apps, navigating menus. Dictation only converts speech to text in a field. People with significant hand pain often use both: Voice Control to move around, and a dictation tool for the writing itself.
Is dictation practical for a whole working day?
For text, largely yes, though your voice gets tired in a way your hands do not, and open-plan offices make it awkward. Most people land on a mixed setup: dictate the bulk of drafting, use hands for short edits and shortcuts, and take voice breaks the same way you would take typing breaks.
Does dictation work for code?
Not well for syntax, and that is fine. Punctuation-dense code is faster to type than to speak. What dictation handles well is everything around the code — comments, commit messages, pull request descriptions, documentation, Slack, tickets — which for many developers is the larger share of daily keystrokes.
Will an employer support voice to text as an accommodation?
Often, yes — input tooling is a common and inexpensive workplace adjustment, and in many jurisdictions employers have a legal duty to consider reasonable accommodations. It is worth asking, and worth asking in writing. Occupational health is usually a better first stop than IT.
Fewer keystrokes, same output
Speak it once, get finished text typed into any Mac app. On-device transcription, nothing to edit afterwards.
Get Started FreeAir Wisper is a native macOS app and requires macOS 14 or later. Free tier includes 200 AI polish requests per week — no credit card.