Your first five minutes with InkSpoke
The goal of this tutorial is simple: go from nothing installed to clean text typed into any app by voice — with a little AI polish — in about five minutes.
What you'll learn
- How to install InkSpoke and complete its short first-run setup.
- How to grant microphone access so the app can hear you.
- How to dictate your first sentence into any app, using push-to-talk.
- How to turn AI refinement on and off, and see the difference it makes.
Prerequisites
- A computer running Windows, macOS, or Linux.
- A working microphone — your laptop's built-in mic is fine.
- An internet connection for the download (dictation itself runs on-device and offline).
That's all. Everything you need to dictate ships with the app, so there's nothing to configure before you start.
This page is the condensed on-ramp. If you'd rather take each step slowly — with every wizard screen and overlay control explained — follow Install and set up InkSpoke and Your first dictation instead. This tutorial covers the same ground, faster.
Time estimate
About 5 minutes, most of it the download and install.
The whole loop, at a glance
Once you're set up, every dictation is the same three-beat gesture: press, speak, press.
Step by step
1. Download and install
Go to inkspoke.app/download and grab the build for your operating system, then install it the usual way for your platform:
| Platform | How to install |
|---|---|
| Windows | Run the installer, then launch InkSpoke from the Start menu. |
| macOS | Move InkSpoke into your Applications folder, then launch it. |
| Linux | Install the downloaded package, then launch InkSpoke. |
The first time you open InkSpoke, it notices setup hasn't run yet and opens the first-run wizard automatically.
2. Walk the first-run wizard
The wizard is a small, centered window with a step indicator and a Skip / Back / Continue footer. It has six steps:
Every step is skippable, but don't skip step 3 — microphone access is what makes dictation work. Click through Welcome and Industry (Industry tunes your workspace vocabulary to your kind of work), then pause at the microphone step:
┌─────────────────────────────────────────────────┐
│ InkSpoke setup │
├─────────────────────────────────────────────────┤
│ │
│ Grant microphone access │
│ InkSpoke needs your mic to hear you speak. │
│ │
│ [ Grant microphone access ] │
│ │
│ ○ ○ ● ○ ○ ○ Step 3 of 6 │
├─────────────────────────────────────────────────┤
│ Skip Back Continue ▸ │
└─────────────────────────────────────────────────┘
Click Grant microphone access, approve the OS prompt, and the button flips to ✓ Granted. Continue through the remaining steps:
- Selected-text capture (step 4) is optional — it powers Command Mode later. Skip it for now if you like.
- Choose your default model (step 5) is informational. It shows what you start with: Whisper Small for speech (on-device, offline, free) and Platform AI for refinement (cloud, part of your Pro trial).
- Ready (step 6) confirms you're set and lists your seeded workspaces. Click Finish.
If you skipped step 3, dictation won't work until you allow the mic. Grant it any time from your OS: macOS System Settings → Privacy & Security → Microphone, Windows Settings → Privacy → Microphone. No reinstall needed.
3. Click into a text field
Open any app with an editable text box — a note, an email draft, a chat window, a plain text editor. Click so the caret is blinking. This is where InkSpoke will type, so make sure it's a real editable field.
4. Press the hotkey and speak
Press the activation hotkey:
| Windows / Linux | macOS |
|---|---|
| Alt + Space | ⌥ + Space |
A small listening overlay appears near the center of your screen, and you'll hear a soft start chime. Now speak one natural sentence — for example, "Thanks for the quick turnaround, I'll review the draft this afternoon." A live waveform ripples as InkSpoke hears you, the label reads Listening…, and a timer counts up.
┌───────────────────────────────────────────────────────────┐
│ ● Listening… 0:03 │
├───────────────────────────────────────────────────────────┤
│ ▁▃▅▇█▇▅▃▂▄▆▇▆▄▂▁▃▅▇▅▃▁▂▄▆▄▂▁ │
│ │
│ [ Workspace ▾ ] [ EN ▾ ] [ Send ] │
├───────────────────────────────────────────────────────────┤
│ ESC cancel · Alt+Space finish 🎙 Your Mic │
└───────────────────────────────────────────────────────────┘
You don't hold the key down while speaking. Press once to start, speak freely, press again to finish. Changed your mind mid-sentence? Press Esc (or click ✕) to cancel — nothing is transcribed or typed.
5. Press the hotkey again to finish
Press the same activation hotkey again (or click Send in the overlay — it does exactly the same thing). The label switches to Processing…, InkSpoke transcribes your speech on-device, and the finished text is typed straight into your field, right where your cursor was. You'll hear a stop chime.
That's the core loop. You just dictated without touching the keyboard.
6. Try AI refinement
By default, InkSpoke lightly polishes what you say — trimming filler, fixing grammar, and matching the tone of the app you're in. You can flip that on and off without opening settings:
| Windows / Linux | macOS |
|---|---|
| Alt + Shift + R | ⌥ + ⇧ + R |
When refinement is off, the overlay shows "AI refinement off" and you get the raw, verbatim transcript. Dictate the same rambling sentence twice — once on, once off — and compare. That difference is the whole point of InkSpoke.
Expected result
You've finished the tutorial when all of these are true:
- Pressing the hotkey pops the overlay and plays a start chime.
- The waveform moves while you speak.
- On the second press, clean, capitalized, punctuated text appears in your field — not a raw string of lowercase words.
- Toggling refinement with Alt + Shift + R visibly changes how polished the output is.
If all four happened, you have the fundamentals. Everything else in InkSpoke builds on this loop.
Troubleshooting
| Problem | Likely cause | Fix |
|---|---|---|
| The overlay never appeared | Another app claimed Alt + Space | Try the secondary hotkey — Ctrl + Shift + Space (⌃ + ⇧ + Space on macOS) — or rebind it in General and hotkeys settings. |
| The waveform stays flat | InkSpoke is listening to the wrong input device | Check the mic name in the overlay footer. Change the input device in Audio and models settings. |
| No text appeared — the overlay shows "No text field detected" | At injection time your cursor wasn't in an editable field | This is the copy-mode fallback protecting your words. Click Copy, then paste where you want it. Next time, click into the text field first so the caret is blinking, then press the hotkey. |
| The text looks raw and unpolished | AI refinement is turned off | Press Alt + Shift + R (⌥ + ⇧ + R) to turn it back on. The overlay confirms the current state. |
Refinement's built-in Platform AI runs as part of a Pro trial that comes with an audio allowance rather than a fixed calendar countdown. See The free trial for what's included.
Next steps
Keep the momentum going:
- Polish your writing with AI — get more out of refinement, from tone to structure.
- Dictate into code and terminals — how InkSpoke behaves in editors and shells.
- Build a workspace for your domain — teach InkSpoke your vocabulary and tone.
Or go deeper on what you just did:
- Push-to-talk basics — the start/stop model, made second nature.
- The listening overlay — every control and state, explained.
- How refinement works — what the AI actually does to your words.
- On-device vs. cloud and privacy — where your speech and text are processed.