Capture, do not compose
Hands-busy dictation earns its place by preventing lost thoughts, not by producing finished writing. The idea you have while stirring a pan is worth thirty seconds of speech and none of the editing you would need to make it publishable.
- One thought per take. Stop when the thought ends.
- Say the conclusion first, in case the rest gets garbled.
- Include enough context to make sense tomorrow.
- Accept rough text. It is raw material.
Trying to compose in these moments is what makes people quit. You cannot review a paragraph with wet hands or on a crowded train, so editing you defer is editing that actually gets done.
The measure of success is how many good thoughts survive to your desk. Rough capture that survives beats a polished note you never started because your hands were full.
Pick the device for the situation
Each situation has a mic problem and a hands problem, and the device that solves both differs more than people expect.
| Situation | Best option | Main obstacle |
|---|---|---|
| Cooking | Watch, or phone propped on the counter | Wet hands, extractor fan, running water |
| Walking outside | Watch or earbuds | Wind and traffic |
| Public transport | Watch, whispering, short takes | Noise and being overheard |
| Driving | Car controls, or wait | Safety comes first |
| Mac docked across the room | Watch as close mic | Distance to the laptop |
The pattern is distance. A mic near your mouth beats a better mic further away in almost every noisy situation, which is why a wrist often wins over a laptop lid.
Set up whichever option fits your most common situation, and stop there. One reliable path you reach for beats four you have to choose between while your hands are covered in flour.
Using a Watch as a close mic
A Watch solves the two hard parts at once: it is already on your wrist and it sits close to your mouth when you raise your arm. Auri Voice supports Apple Watch dictation for exactly these cases.
- Loud environments, where distance to a laptop mic is the problem.
- Commutes, where you are moving and the phone is packed away.
- Whispering, where being close matters more than anything else.
- A docked Mac, when the machine is across the room from you.
Audio from the Watch is encrypted in transit and transcribed locally on the Mac, and the text arrives on the laptop while it is open. Recognition stays on the machine you own rather than moving to a server.
Raise your wrist to a comfortable talking distance rather than holding it at your chin. A steady position through the whole take matters more than being very close for the first three words.
Keep takes short and label them
The reason captures go unused is that nobody can face a list of twelve unlabeled paragraphs. Two words at the start of each take fixes that.
- Start with a tag: groceries, blog idea, reply to Sam, bug.
- Say the point in one sentence.
- Add the one detail you would forget.
- Stop. Start a new take for the next thought.
Tags make triage skimmable and let you batch. Six takes tagged the same way get handled in one pass at the desk instead of six context switches.
When you lose the thread mid-take, stop and start again. Talking your way back leaves both versions in the text, and untangling that later costs more than re-speaking two sentences.
Noise, wind, and whispering
Environment does more damage than any setting, and small physical adjustments beat configuration changes every time.
- Turn away from wind, or shield the mic with your hand.
- Pause for the truck, the blender, the train announcement. Two seconds costs nothing.
- Turn off the radio or TV. Background speech is the worst kind of noise.
- Whispering: get closer, slow down, keep takes under fifteen seconds.
- Running water and extractor fans are loud in a way you stop noticing. They are still loud.
Whispering deserves an honest expectation. It removes vocal cord vibration that models rely on, so accuracy drops even with a close mic. It still beats losing the thought on a quiet train.
In loud places, slow down slightly rather than raising your volume. Louder speech distorts more, while a steady moderate pace with clear sentence endings gives the engine something to work with.
The five-minute cleanup at your desk
Capture is worthless without a pass that closes the loop. Five minutes, once a day, and most items should end up deleted.
- Read the list top to bottom without editing anything.
- Sort into three piles: act now, keep for a project, delete.
- Handle the act-now items immediately, since most are two-line replies.
- Move project material into the document where it belongs.
- Delete the rest without ceremony.
The delete pile is the important one. A capture habit without a delete habit turns into a backlog you feel guilty about, and the guilt is what makes people stop capturing.
Doing this at the same point each day, at lunch or before you close the laptop, keeps the list short enough that reading it never feels like a chore.
Limits and manners
A few limits are worth respecting, and they have less to do with software than with where you are standing.
- Driving: safety first. Use car controls or wait for a stop.
- Public spaces: your half of a work conversation carries further than you think.
- Confidential material: a train is not the place, whispering or not.
- Other people: dictation is for your own speech, not for capturing others.
- Battery: long sessions on a Watch or phone in the cold drain faster.
The overheard risk is the one people underestimate. Client names, salary numbers, and health details spoken on a platform reach whoever is standing there, and no local processing helps with that.
When the content is sensitive, keep the capture generic. A note that says call back about the March issue is enough for you and useless to anyone nearby.
FAQ
How long should a hands-busy take be?
Twenty to forty seconds, one thought each. Long takes in noisy places give the engine more chances to go wrong in a row, and one bad guess drags the rest of the sentence with it.
Is a Watch better than a phone?
For hands-busy moments, often yes, because your wrist is closer to your mouth than a phone in a pocket or a laptop across the room. Auri Voice supports Apple Watch dictation for loud rooms, commutes, whispering, and a docked Mac.
Does whispering work?
It works well enough for capture and worse than normal speech, because whispering removes the voicing that speech models rely on. Get the mic close, speak slowly, keep it short, and expect more cleanup.
What about dictating while driving?
Driving safety comes first, and your car’s own controls exist for this. Do not reach for a phone or fiddle with a Watch while moving, and let the passenger seat or a parked moment handle anything longer than a sentence.
How do I stop the notes from piling up?
Give every take a two-word label as you speak, then run a five-minute triage at your desk. Most captures should be deleted after you act on them. A capture habit without a delete habit becomes a graveyard.