Voice-first writing,
built honestly.
We're a small team that thinks voice is the most natural way to put words on a screen. InkSpoke is what we wished existed.
Close the gap between what you say and what you'd actually want to send.
We believe voice is the most natural way to communicate ideas — yet most voice-to-text tools produce raw, unformatted transcriptions that require extensive editing. InkSpoke exists to close that gap.
InkSpoke is an app you install. It converts your speech into polished, context-aware text and injects it directly into whatever application you are working in. No browser tabs, no copy-paste, no formatting cleanup.
The storyStarted as a frustration, ended as a product.
InkSpoke started as a personal frustration: existing dictation tools either produced sloppy output, required cloud processing for every word, or were locked to a single platform. We wanted something that worked offline, respected privacy, and understood context.
Built on Whisper for transcription and local LLMs for refinement, InkSpoke runs entirely on your device when you want it to. Plug in your own API keys to cloud providers and you stay in control — your terms, your keys, your cost.
Three things, in this order.
Privacy first
Your voice, your data, your device. InkSpoke processes everything locally by default. We never store your audio or transcriptions on our servers.
Speed matters
Every millisecond counts when you are in flow. Our end-to-end latency budget is under 600ms from hotkey press to text injected — fast enough to feel instant.
Cross-platform, properly
One app that works the same on macOS, Windows, and Linux. No second-class platforms, no feature gaps. Native performance everywhere.
You spoke. Now ink it.
Free with platform-hosted voice + text models. Pro unlocks BYOK, on-device models, and unlimited workspaces.