Push-to-talk basics
Push-to-talk is the heart of InkSpoke: press a hotkey, speak, press again, and your words land at your cursor. This page walks through that loop and every control that drives it — the activation and secondary hotkeys, the Send button, canceling, the sound cues, and the safety timeouts that stop a runaway recording.
The basic loop
One hotkey does almost everything. You press it to start listening, and you press the same hotkey again to finish — InkSpoke then transcribes what you said, optionally refines it with AI, and injects the result into whatever app had focus.
- Press the activation hotkey — default Alt + Space on Windows and Linux, or ⌥ + Space on macOS. The listening overlay appears.
- Speak. A live waveform shows InkSpoke is hearing you, and an elapsed-time counter ticks up.
- Press the hotkey again (or click Send) to finish. InkSpoke transcribes, refines if enabled, and injects the text.
Under the hood, each dictation moves through a small state machine:
The controls at a glance
While you're listening, the overlay gives you everything you need:
┌────────────────────────────────────────────────────────┐
│ ● Listening… 0:04 ✕ │
├────────────────────────────────────────────────────────┤
│ ▁▃▅▇▅▃▁▂▄▆▄▂▅▇▅▃▁ "your words appear here" │
│ │
│ [ Workspace ▾ ] [ EN ▾ ] [ Send ] │
├────────────────────────────────────────────────────────┤
│ ESC cancel · Alt+Space finish 🎙 Microphone │
└────────────────────────────────────────────────────────┘
- The activation hotkey starts and finishes the dictation. Pressing it while InkSpoke is listening is the same as clicking Send.
- Send stops recording and processes exactly like a second hotkey press — handy when your hand is already on the mouse.
- Esc or the ✕ button cancels: the session is thrown away with no transcription and nothing is injected.
- The footer always reminds you of both: "ESC cancel · Alt+Space finish", along with the microphone InkSpoke is listening on.
For a full tour of the waveform, pickers, and the different overlay looks, see The listening overlay.
Starting and stopping
Any of these will finish a dictation and kick off processing:
- Pressing the activation hotkey a second time.
- Pressing the secondary activation hotkey (see below) — it behaves identically.
- Clicking Send on the overlay.
There's a second, fully independent hotkey that does the exact same start/stop as the primary one. By default it's Ctrl + Shift + Space on Windows and Linux, or ⌃ + ⇧ + Space on macOS. It's there so you have a fallback if the default Alt + Space clashes with something else on your system.
Keep a hand on the keyboard? Use the hotkey to both start and stop. Already reaching for the mouse? Click Send to finish. They do the same thing — use whichever is faster in the moment.
Canceling a dictation
To abandon a recording without transcribing anything, press Esc or click the ✕ on the overlay. While the overlay is open, Esc is registered as a global shortcut, so it works even though the overlay isn't the focused window.
Canceling throws the audio away — there's no transcript, no refinement, and nothing is injected or saved to history. If you want the text, use Send (or the hotkey) instead.
Sound cues
InkSpoke plays short chimes so you can dictate without watching the screen:
| Cue | When it plays |
|---|---|
| Start | Listening begins |
| Stop | You finish and processing starts |
| Cancel | You abandon the session |
| Error | Something in the pipeline fails |
All four are controlled by a single toggle. Turn them off if you'd rather work in silence.
| Setting | Default | What it does |
|---|---|---|
SoundEffectsEnabled | On | Plays the start / stop / cancel / error chimes during capture |
Safety timeouts
Two limits make sure a forgotten overlay never sits listening forever, and that a single recording can't grow without bound. Both can be set to 0 to disable them entirely.
| Setting | Default | What it does |
|---|---|---|
SilenceTimeoutSeconds | 30 | Cancels the session automatically after this many seconds of continuous silence (0 = off) |
MaxRecordingDurationSeconds | 300 | The longest a single recording can run before InkSpoke stops it — 5 minutes by default (0 = no limit) |
The silence timeout counts continuous silence — as long as InkSpoke keeps detecting speech, the clock resets. It's there to end a session you walked away from, not to cut you off mid-sentence.
Hotkeys by platform
Default bindings differ only in how the keys are written — macOS uses the ⌥ ⌃ ⇧ glyphs, while Windows and Linux spell them out.
| Action | Windows / Linux | macOS |
|---|---|---|
| Start / stop dictation (activation) | Alt + Space | ⌥ + Space |
| Start / stop dictation (secondary) | Ctrl + Shift + Space | ⌃ + ⇧ + Space |
| Send (stop & process) | Send button or the activation hotkey | Send button or the activation hotkey |
| Cancel | Esc or ✕ | Esc or ✕ |
Both activation hotkeys are configurable. Esc to cancel is fixed and can't be remapped. To rebind the activation or secondary hotkey — or to see the other InkSpoke shortcuts — head to Settings → General & hotkeys.
If Alt + Space collides with another app (on Windows it opens a window's system menu), you don't have to rebind anything — just reach for the secondary Ctrl + Shift + Space, which triggers the identical start/stop. InkSpoke also warns you if a hotkey you choose conflicts with a reserved OS combo.
Next steps
- The listening overlay — the waveform, pickers, and overlay styles in detail.
- Dictation modes and languages — Standard vs. Live Preview, and dictating in other languages.
- How dictation works — the full capture → transcribe → refine → inject pipeline.
- Settings → General & hotkeys — rebind the activation and secondary hotkeys.