Hello world VoxWave is live!
After 6 months of solo dev, VoxWave is publicly available on Windows and Linux.
It's a desktop dictation tool: press a hotkey, speak, and clean text gets pasted wherever your cursor is. Free, works offline (local Whisper) or in the cloud (Groq) — your choice. I built this because I didn't want to pay 20 USD for Wispr Flow.... Looking for the first 100 users and would love brutal feedback.
Solid call on Groq for the cloud path, latency is make-or-break for hotkey dictation flow and Groq smokes most alternatives there. The local Whisper fallback is the right escape hatch for offline.
One question for the first 100: which Whisper model size are you shipping locally? Tiny gets fast first paste but accuracy drops hard on technical vocab. Medium feels like the sweet spot, but the download alone gates a lot of casual users from the offline mode.
We ship
small(244M) by default same trade-off you flagged:basewas breaking on language auto-detect with technical vocab (caught it flipping French to Russian on 3-4s clips), andmediumkills the hotkey UX without a GPU. Model isn't bundled installer stays at 105MB,smallauto-downloads on first use of local mode so casual cloud users never see it. Power users can override in Settings orconfig.yaml. Are you using local mode yourself? On what hardware?