Set up InkSpoke on Android
This page gets InkSpoke running on your Android phone: install the app, complete the short onboarding wizard, grant the permissions it needs, and — the one step people miss — enable the InkSpoke voice keyboard so you can dictate system-wide, into any app.
InkSpoke on Android is voice-keyboard-first. You dictate two ways: inside the app on the Home screen (great for capturing a quick note you'll copy elsewhere), and through the InkSpoke voice keyboard, which drops a mic into any text field on your phone. Setup takes about five minutes.
Before you start
| Requirement | Detail |
|---|---|
| Android version | Android 8.0 (API 26) or newer. |
| A microphone | The phone mic works; Bluetooth, wired, and USB mics are detected too. |
| Internet (for the default model) | The default speech model is the cloud Platform model, so you'll want a connection for first use. You can switch to a fully offline on-device model later. |
Step 1 — Install the app
Get InkSpoke from Google Play (search for "InkSpoke"), or start from the InkSpoke download page. Open the app once it finishes installing.
Step 2 — Walk through onboarding
The first time you launch InkSpoke, a 6-page setup wizard appears before the main app. Swipe through it, or tap Skip (top-right) to jump straight to the app and finish setup later.
┌─────────────────────────────────────────────────┐
│ [Skip] │
│ │
│ Welcome to InkSpoke │
│ │
│ you spoke, it's inked. │
│ │
│ │
│ ● ○ ○ ○ ○ ○ │
│ [ Next ] │
└─────────────────────────────────────────────────┘
Here's what each page does:
- Welcome — a quick intro to what InkSpoke does.
- Industry — "What kind of work do you do?" Pick a preset chip, or choose Other… and type your own. Your answer seeds your workspace vocabulary so transcription and refinement fit your field from day one.
- Microphone — tap Allow Microphone to grant recording access. See Step 3 for the details.
- Selected-text capture (Command Mode) — explains a feature that lets you select text in another app and speak an edit to it. On Android this needs an Accessibility service that isn't available yet, so this page is informational only — see the note below.
- Model picker — choose your default speech model. The built-in Platform (cloud) model is selected for you. You can switch to an offline on-device Whisper model or your own provider afterward in Settings → AI Models.
- Ready — a recap that lists the starter workspaces InkSpoke sets up for you (Speech / Voice, Chat, Email, Note Taking). Tap Start using InkSpoke to finish.
The onboarding page about selecting text and speaking an edit (Command Mode) describes a feature that is not yet available on Android — it depends on an Accessibility service that hasn't shipped. Treat that page as a preview of what's planned, not something you can turn on today. Dictation into any app still works through the voice keyboard.
After the wizard you land on the Home tab — the in-app dictation screen. The bottom navigation gives you Home, History, Workspaces, and Settings.
Step 3 — Grant permissions
InkSpoke asks for two Android permissions. Both are needed for smooth, system-wide dictation.
| Permission | When you're asked | Why InkSpoke needs it |
|---|---|---|
| Microphone | On the Microphone onboarding page, and again the first time you tap record | To capture your speech for transcription. Without it, the keyboard mic shows an error state and can't record. |
| Notifications (Android 13+) | The first time the background model service starts | So InkSpoke can show its "InkSpoke Active" service notification while the speech model is loaded. |
The "InkSpoke Active" notification is expected
To keep dictation responsive across apps, InkSpoke runs a small foreground service that hosts the speech-recognition model for the keyboard. While it's running you'll see a persistent notification such as "InkSpoke Active" ("Speech recognition model loaded" or "Loading model…"). This is normal — leave it in place. If you force-stop or dismiss the service, the keyboard's mic may need to reload the model before your next dictation.
Step 4 — Enable the InkSpoke voice keyboard
This is the step that unlocks dictation everywhere. Android treats the InkSpoke keyboard as an input method (IME) that you have to switch on yourself, then select when you want it.
- Open the Android Settings app.
- Go to System → Languages & input → On-screen keyboards (the exact wording varies a little by phone brand — look for the list of installed keyboards).
- Turn on InkSpoke. Android may warn that a keyboard can read what you type — that's the standard IME prompt; confirm to enable it.
- Open any app with a text field and tap it so the keyboard appears.
- Use your phone's keyboard-switcher (often a small keyboard icon in the navigation bar, or the globe key) and choose InkSpoke.
Once InkSpoke is the active keyboard, you'll see its compact voice toolbar instead of a normal typing keyboard:
┌─────────────────────────────────────────────────┐
│ [ Workspace ] EN ⌨ switch │
│ │
│ ( ◉ ) │
│ tap to dictate │
└────────────────── ───────────────────────────────┘
Tap the big mic to start and stop; your words are transcribed and dropped straight into the field you were typing in. The switch keyboard button on the toolbar takes you back to your previous keyboard. Full details are in The InkSpoke voice keyboard.
The InkSpoke keyboard is a voice toolbar, not a full typing keyboard — there are no letter keys or word suggestions. Use it to dictate, then switch back to your usual keyboard for typing. The toolbar's language and workspace chips are display-only for now; you set your language and workspace in the main app (see Dictation and features).
Run onboarding again
Want to revisit the wizard — for example to change your industry or re-pick a model? You can replay it any time:
Settings → Onboarding → Re-run onboarding → Run again
This walks you back through all six pages without reinstalling the app.
Power-user notes
- Start offline. Prefer to keep audio on your device? After setup, go to Settings → AI Models → On-Device, download a Whisper model, and set it as your default. Downloads are restricted to Wi-Fi by default. Once a model is downloaded, dictation runs fully on-device.
- Bring your own provider. In Settings → AI Providers you can add an OpenAI-compatible provider (base URL + API key) and use it for transcription and/or AI refinement.
- Tune what InkSpoke knows about you. Settings → Personal Context and Settings → Vocabulary let you add an "about me" blurb and a term list that bias recognition and refinement — worth a minute if you use lots of jargon or names.
Next steps
- The InkSpoke voice keyboard — how the voice toolbar works and how to dictate into any app.
- Dictation and features on Android — models, languages, workspaces, and history.
- All mobile apps — how iOS and Android compare and what's coming next.