Go hands-free with voice control
The goal of this tutorial is to put down the keyboard entirely: you'll turn on InkSpoke's spoken wake word, dictate a sentence start-to-finish by voice, and — if you're on Pro — add a ready-made voice command and run it, all without touching a key.
What you'll learn
- How to enable the offline wake word and download its model.
- How to pick your wake name (and an optional greeting prefix).
- How to start, finish, and cancel a dictation entirely by voice.
- How to submit hands-free with the quick-confirm phrase.
- (Pro) How to add a voice command from the template catalog and fire it.
Prerequisites
- InkSpoke desktop installed and set up, with microphone access granted. If you haven't done that yet, start with Install and set up InkSpoke.
- You've done at least one push-to-talk dictation, so the listening overlay is familiar. If not, walk through Your first dictation first.
- A quiet-ish spot and a working mic — the wake word listens continuously in the background.
- For the voice-command steps only: a Pro (or Perpetual) plan. You can build commands on any plan, but they won't actually fire until you're on Pro. See Plans and pricing.
InkSpoke's wake-word engine recognizes English only. You'll pick an English name, and the wake/finish/cancel phrases are matched against an English speech model. There's no non-English wake word today.
Time estimate
About 10 minutes — plus a one-time model download (about 42 MB) the first time you enable the wake word.
The hands-free loop, at a glance
Once the wake word is on, a whole dictation happens by voice: you say a name to start, speak, then say a phrase to finish.
Step by step
1. Enable the wake word
Open Settings → Hands-Free, then select the Wake Word tab and toggle Enable Wake Word.
The first time you enable it, InkSpoke asks to download the offline wake-word model (about 42 MB) and shows a consent dialog. Approve it — the download runs once, and after that everything works offline.
┌─ Settings › Hands-Free ─────────────────────────────────────┐
│ [ Overview ] [ Wake Word ] [ Voice Commands PRO ] │
├─────────────────────────────────────────────────────────────┤
│ Enable Wake Word .................................... (●) │
│ │
│ Wake name [ Annika ▾ ] Prefix [ (none) ▾ ] │
│ ▶ Preview │
│ │
│ Say to start "annika" │
│ Say to finish "thank you annika" │
│ Say to cancel "cancel annika" │
│ │
│ Detection sensitivity ............... [ Very High ▾ ] │
│ Quick confirm ....................... (●) │
└─────────────────────────────────────────────────────────────┘
The Overview tab (the first sub-tab) gives you an at-a-glance summary of everything InkSpoke is currently listening for, if you'd like to confirm what's active.
Audio for wake detection never leaves your machine — the model runs locally. See Wake word setup for how to scope when it listens (schedules, idle/sleep auto-disable, pause-on-music).
2. Pick your wake name
InkSpoke ships with a curated list of names the offline model recognizes reliably. Pick one from the Wake name picker — the default is Annika — then click Apply (it only appears once you've changed something).
| Name | Name |
|---|---|
| Annika (default) | Samantha |
| Athena | Serena |
| Cassandra | Stella |
| Miranda | Victoria |
| Rebecca | Zeno |
Click ▶ Preview to hear the wake phrase spoken aloud by your operating system's text-to-speech voice, so you know what to say. (If your system has no TTS voice, the preview is quietly skipped.)
Whatever name you choose, the finish and cancel phrases follow automatically: "thank you <name>" to finish, "cancel <name>" to abort.
A single short word can occasionally trigger by accident. Set the Prefix to Hey or Hi to give the listener a longer, more distinctive phrase — your start phrase becomes "hey annika" or "hi annika." The finish and cancel phrases stay the same. Pro users can also type a fully custom name via Custom… in the picker; see Wake word setup.
3. Dictate hands-free: wake, speak, finish
Click into any editable text field — a note, an email draft, a chat box — so the caret is blinking. That's where your words will land. Then, without touching the keyboard:
- Say your wake name — "annika." The listening overlay opens, exactly as if you'd pressed the push-to-talk hotkey.
- Speak your sentence naturally. The waveform ripples as InkSpoke hears you.
- Say the finish phrase — "thank you annika." InkSpoke transcribes your speech, refines it (if refinement is on), and types the result at your cursor.
Everything is case-insensitive. Changed your mind mid-sentence? Say the cancel phrase — "cancel annika" — and the recording is thrown away without transcribing.
4. Submit hands-free with quick-confirm
Quick confirm is on by default, and it's what makes a dictation fully hands-free: after your text is injected, InkSpoke opens a short window during which a confirm phrase presses Enter for you — so you can proofread, then send, without reaching for the keyboard.
- The default confirm phrase is "ok confirm" (both "ok" and "okay" count).
- You say it without the wake name — it's a follow-up to the dictation you just finished.
- The default window is 60 seconds (adjustable 1–300 in the Wake Word tab).
So a complete hands-free round trip into a chat box sounds like:
"annika" … "See you at three, I'll bring the deck." … "thank you annika" … "ok confirm"
…and your message is typed and sent, keyboard untouched.
5. (Pro) Add a voice command from the template catalog
Voice commands let you do more than dictate — you can open apps, press keys, and insert text by voice. This is a Pro feature: you can build commands on any plan, but they only fire on Pro or Perpetual.
- Still in Settings → Hands-Free, switch to the Voice Commands tab (marked with a PRO badge) and turn on Voice Commands. (The wake word must be enabled — commands ride on the same listener.)
- Click Manage Mappings, then Browse Templates to open the searchable catalog of ready-made commands, grouped by category.
- Find a simple, observable one to start with — the Web category's "open google" is ideal, since running it visibly opens your browser. Add it directly, or choose Edit before adding to open the normal form pre-filled.
┌─ Browse Templates ──────────────────────────────────── ──┐
│ Search [ open google ] │
│ Category [ Web ▾ ] │
├─────────────────────────────────────────────────────────┤
│ 🔗 open google Opens google.com in your browser │
│ [ + Add ] │
│ 🔗 open youtube Opens youtube.com │
│ 🔗 open github Opens github.com │
│ 🔗 open chat gpt Opens chatgpt.com │
└─────────────────────────────────────────────────────────┘
The catalog spans Terminal, App Controls, Text Autofill, Web, Editor, Productivity, and History categories, so once you're comfortable you can add commands like "git status," "open {app}," or "insert signature." Only "insert last item" is pre-installed — everything else you add yourself.
6. (Pro) Run the command
Every voice command follows the same shape: your wake name, the reserved word action, then the phrase you added. So with the default name and the "open google" template:
"annika action open google"
Saying action is what tells InkSpoke a command is coming (rather than a dictation); it opens a short (about 5-second) window to catch your phrase. Say the phrase in that window, and your default browser opens google.com.
Continuous Commands is on by default, so after a command runs the listener stays armed for a short follow-up window — you can fire the next command without saying the wake name again. See Voice commands for the full command reference.
Expected result
You've finished when all of these are true:
- Saying your wake name pops the listening overlay, just like the push-to-talk hotkey.
- A full wake → speak → finish cycle types clean text at your cursor without a keypress.
- Saying "cancel <name>" during a dictation throws it away with nothing typed.
- The quick-confirm phrase ("ok confirm") presses Enter for you after text lands.
- (Pro) "<name> action open google" opens google.com in your browser.
Troubleshooting
| Problem | Likely cause | Fix |
|---|---|---|
| InkSpoke never reacts to my wake name | The word wasn't recognized, or sensitivity is too low | Say the name clearly and on its own. If it's still missed, nudge Detection sensitivity up (default is already Very High), or add a Hey / Hi prefix for a more distinctive phrase. Confirm the wake word is enabled on the Overview tab. |
| It keeps waking up on its own | Sensitivity is high, or conversation/media is triggering it | Drop Detection sensitivity to High or Normal, keep Pause on music playback on, and consider a prefix. See Wake word setup for schedules and idle auto-disable. |
| My spoken command does nothing | Voice Commands is off, or you're not on Pro | Voice commands require Pro/Perpetual and the wake word enabled. If InkSpoke hears a mapped command while Voice Commands is off, it shows a balloon offering to turn them on. Check the Voice Commands tab. |
| The wrong phrase gets caught after "action" | Long or unusual phrases are hard for the offline engine | Keep phrases short and predictable, and start from the template catalog. Teach the engine your app names and jargon via Favorites and vocabulary. |
Next steps
- Wake word setup — tune sensitivity, schedules, quick-confirm, and continuous-recording-after-wake.
- Voice commands — the full Pro reference: all ten action types, placeholders, the built-in press command, and bundle export/import.
- Favorites and vocabulary — add your apps and domain words so commands resolve reliably.
- Command Mode — transform selected text by voice, triggerable hands-free with "<name> command."
- Your first five minutes — the push-to-talk fundamentals this tutorial builds on.