One Shortcut, Three Modes: Designing Voice Input
Good voice input on a Mac is not about memorizing a dozen commands. It is about one shortcut that always works, split into a few clear modes you can reach without thinking. Here is how to design that setup so speaking feels as fast as your cursor.
Key takeaways
- One global shortcut that works system-wide beats a menu you have to hunt through.
- Three modes cover almost everything: raw dictation, AI cleanup, and translate-as-you-speak.
- Every mode runs on-device, so your audio and transcript never leave the Mac.
- Choose toggle or push-to-talk based on whether you dictate in long passages or short bursts.
Why one shortcut beats a busy menu
The whole point of voice input is speed. Most people speak around three to four times faster than they type, so any friction in the trigger cancels out the gain. If starting dictation means clicking an icon, waiting for a panel, or picking an option from a list, you have already lost the advantage.
The fix is a single global shortcut that behaves the same everywhere. Because BlaBlaType works system-wide, the same key press dictates into email, Slack, Notion, a code editor, or an AI chat. There is no per-app setup and no window to bring forward. You keep your hands on the keyboard, tap the shortcut, and talk. For people managing repetitive strain from heavy typing, cutting that friction matters, and health bodies like the NHS guidance on RSI point to reducing keyboard load as one practical step.
The three modes, and what each is for
One shortcut is the anchor. The design work is deciding what happens after you speak. Three modes cover the vast majority of real writing, and each maps to a clear intent.
- Raw dictation. Your words land close to verbatim. This is the fastest path and the right choice for quick notes, chat replies, code comments, or anywhere you want exactly what you said.
- AI cleanup. The same speech is rewritten on-device: filler words removed, punctuation and grammar fixed, tone adapted. This is for messages and documents you want to send without a second edit.
- Translate-as-you-speak. You talk in one language and the text arrives in another. This is the mode for bilingual teams and cross-border email.
| Mode | Best for | Output | On-device |
|---|---|---|---|
| Raw dictation | Notes, chat, code | Close to verbatim | Yes |
| AI cleanup | Email, docs, posts | Polished, ready to send | Yes |
| Translate-as-you-speak | Cross-language writing | Text in the target language | Yes |
The key detail: all three run locally. Speech recognition happens on your Mac with local Whisper and Parakeet models, and the AI cleanup step is powered by Apple Intelligence on-device. If you want the mechanics, we cover how on-device AI rewrites text without the cloud in a separate guide.
How to set it up in three steps
You do not need a complex configuration. The setup is short, and once it is done the shortcut is muscle memory within a day.
Pick one shortcut you will never fat-finger
Choose a key combination that does not clash with your apps. Decide between a toggle, which starts and stops on separate presses, and push-to-talk, which records while you hold the key.
Assign a second shortcut for cleanup
Keep raw dictation on your main key, then map a nearby second shortcut to the AI cleanup mode. Now one hand position gives you both verbatim and polished output without opening a menu.
Teach it your words, then flip on translate when needed
Add names and jargon to the custom dictionary so transcription nails them, and toggle translate-as-you-speak only for the languages you actually write in.
What happens the instant you press the key
Understanding the pipeline helps you trust it. From the moment you tap the shortcut, everything below runs on your machine. Nothing is uploaded, and there is no round trip to a server that could stall or leak.
Because the audio and transcript never leave the device, this design suits sensitive work by default. That privacy floor is why teams handling client or clinical notes lean on local dictation, a point we explore in whether Mac dictation is actually private.
Set up your one shortcut
Dictate, clean up, or translate with a single key press. Everything stays on your Mac. No card needed for the trial.
Download for macOSToggle or push-to-talk, and who each mode suits
The last design choice is how the key behaves. A toggle starts recording on the first press and stops on the second, which is ideal for long passages where you do not want to hold anything. Push-to-talk records only while the key is held, which suits short bursts and keeps stray speech out of your text. Neither is better in the abstract; pick the one that matches how you write.
The three modes flex across very different users. A writer lives in AI cleanup, turning loose spoken drafts into publishable prose. A developer stays in raw dictation for comments and commit messages, where verbatim beats polish. People who find sustained typing draining, including many in the ADHD community served by CHADD, use the shortcut to get thoughts down before they slip. If you are weighing this against the built-in option, our Apple Dictation vs BlaBlaType comparison lays out where a dedicated tool pulls ahead, and you can see plan details on the pricing page.
Frequently asked questions
How many shortcuts do I need for voice input?
One is enough to start. A single global shortcut can trigger recording anywhere on your Mac. If you want fast switching, add a second shortcut so one key gives you raw dictation and the other adds AI cleanup, without opening any menu.
What is the difference between raw dictation and AI cleanup mode?
Raw dictation types your words close to verbatim, which is fastest for notes and code. AI cleanup takes the same speech and rewrites it on-device: it removes filler words, fixes punctuation and grammar, and can adapt tone, so the text is ready to send.
Does a translate mode send my audio to the cloud?
No. BlaBlaType runs speech recognition on your Mac, and its optional translate-as-you-speak feature works within that on-device pipeline. Your audio and transcript never leave the device, so translation does not require uploading your voice to a server.
Can I use push-to-talk instead of a toggle?
Yes. You can configure the shortcut as a toggle that starts and stops recording, or as push-to-talk where you hold the key while you speak and release to finish. Push-to-talk suits short bursts, while a toggle suits longer dictation.
Will one shortcut work in every app?
Yes. BlaBlaType works system-wide, so the same shortcut dictates into any app or text field: email, Slack, Notion, a code editor, or an AI chat. The text lands wherever your cursor is, with no per-app setup.