The thought is fully formed for about three seconds. Then you reach for the keyboard, click into the right window, remember you needed to open a different app first, and by the time the cursor is blinking the thought is gone. You know it was good. You can't get it back.
If that loop is familiar, the problem usually isn't the idea. It's the gap between having it and getting it down. Speaking closes that gap. This is a practical guide to setting up voice to text for ADHD on a Mac — what actually helps, how to configure it, and where it doesn't help so you don't waste a week finding out.
Quick caveat before anything else: this is a workflow guide from people who build a dictation app, not medical advice. It's about the mechanics of capturing your own words faster.
Why speaking helps when typing fights you
Three things line up.
Speech keeps pace with the thought. Most people speak at 130–150 words per minute and type at 40–60. When ideas arrive faster than you can record them, that 3x difference is the margin between catching one and losing it. You don't have to hold the thought in your head while your hands catch up.
It sidesteps the blank-page freeze. Starting is often the hardest part — not knowing what to say, but committing the first word to an empty document. Talking has a lower activation cost than typing. You can start a sentence out loud before you've decided how it ends, the same way you would if a friend asked what you were working on.
It offloads working memory. Holding a task in your head and operating the keyboard and not getting pulled into the six other tabs is a lot of simultaneous load. Saying the thing out loud gets it out of your head and onto the screen in one move, which frees up the attention you were spending just to keep it from slipping away.
The real catch: capturing is easy, organizing is hard
Here's the honest part. Voice makes capture almost frictionless — which means it's easy to end up with a pile of captured thoughts and no structure. Trying to dictate perfectly organized, tagged, prioritized notes in one pass adds exactly the kind of friction voice was supposed to remove.
The pattern that works: capture fast, organize later. Get the thought down the second you have it, in whatever form. Sort it during a separate, scheduled pass — not in the moment. Two different jobs, two different times. Fighting to do both at once is where most setups fall apart.
What you need on a Mac
- An Apple Silicon Mac (M1 or newer)
- A decent mic — the built-in one is fine to start; AirPods or a headset are better
- A dictation app that can capture a quick thought into any app without you switching windows
- A single place the captured thoughts land
That third point matters most. If capturing requires opening a specific app first, the friction is back and you'll stop doing it.
Apps worth knowing about
Apple Dictation. Free, on-device, already on your Mac. Toggle it in System Settings and it types into any field. Good for a quick sentence. The limits: it stops after short bursts, it transcribes filler and false starts literally, and there's no capture history — dictate a thought into the wrong window and it's just gone.
Dragon. You'll still see it recommended for ADHD. Worth knowing that Nuance discontinued the Mac version years ago — the current Dragon is Windows-only. If you're on a Mac, those recommendations are dated.
Willow, Wispr Flow, and similar cloud tools. They work system-wide and are polished. The trade-offs are a monthly subscription (Wispr Flow is around $15/month) and that your audio is processed on their servers. Fine for many people; a dealbreaker if you'd rather nothing leave your machine.
Vext — $49 once, Apple Silicon only, fully local. Built around exactly the capture-fast pattern: a hands-free toggle so you press once and talk, an Enhance pass that cleans the ramble into readable text, and a Notes area where captures land. No account, no subscription, nothing sent to a server.
Superwhisper, MacWhisper. Both solid local options too. Superwhisper is subscription or a steep one-time tier; MacWhisper (€64 Pro) leans toward transcribing recordings. Any of these three can do the job — the thing to check is how little effort it takes to fire off a capture.
Setting up Vext for fast capture
The configuration built for this:
- Install:
brew install muvon/tap/vext - Grant Microphone and Accessibility permissions when asked
- Open Settings > Hotkeys and set a hands-free hotkey — press once to start, press once to stop, no holding a key while you talk
- Turn on Enhance (default Gemma 3 4B model) so the messy capture comes out readable
- Keep YOLO Mode off — you don't want anything auto-submitting a half-formed thought
The hands-free toggle is the piece that matters here. Holding a key down while a thought pours out is its own small distraction; a single press to start and stop gets out of the way.
The capture-then-sort workflow
In the moment: thought shows up, press the hotkey, say it, press again. Don't clean it up, don't categorize it, don't decide where it goes. Two seconds, done, back to what you were doing.
Later, on a schedule: once a day or twice a week, open the pile and sort. Move things into your task manager, your notes app, your calendar. This is the boring pass, and doing it separately is the whole trick — you're not asking the distractible moment to also be the organized one.
Enhance earns its keep here because voice captures are rambly by nature. A 40-second brain-dump full of "um, so, the thing is" comes back as a couple of clean sentences you can actually read a week later, with the raw version saved underneath in case the cleanup dropped something.
A few things that help
- One thought per capture. Resist chaining five ideas into one recording. Short captures are easier to sort later.
- Don't self-edit while speaking. If it comes out wrong, just say it again correctly and keep going. The cleanup pass usually drops the false start. Stopping to fix yourself mid-sentence is where the momentum dies.
- Say it before you judge it. The point of fast capture is to beat the internal filter that talks you out of ideas. Get it down first, decide if it was any good during the sorting pass.
- Put the pile somewhere you'll actually reopen. A capture you never revisit is just a nicer way of forgetting.
Where voice doesn't fix anything
Being straight about this saves you disappointment.
Voice captures thoughts. It does not organize them, prioritize them, or make you do them. The executive-function part — deciding what matters and following through — is still yours. If the hard part for you is starting a task you've already written down, faster capture won't touch that.
It also won't help in a room where you can't talk out loud, and it won't make a noisy café workable without decent headphones. And like any new input habit, it feels awkward for a couple of weeks before it feels natural. Give it that long before deciding.
What it does do, reliably, is shrink the distance between having a thought and having it saved. For a brain that loses good ideas in that exact gap, that alone is worth the setup.
Try Vext — the free trial includes 100 dictations and 50 notes, no account and no credit card, so you can find out in an afternoon whether talking-to-capture fits how your head works.