VoiceInk is the closest philosophical relative to OpenWispr — GPL-licensed, on-device via whisper.cpp and Parakeet, buildable free from source. It's Apple Silicon macOS only, and has the best transcription-recovery UX found anywhere in this research.
VoiceInk (github.com/Beingpax/VoiceInk) runs on-device first, via whisper.cpp and Parakeet through Apple's FluidAudio — the same engine family OpenWispr uses. It's GPL-3.0 licensed and free to build from source; paying unlocks auto-updates and support, as a one-time purchase rather than a subscription. It's macOS, and specifically Apple Silicon only — no Intel Mac, no Windows, no Android.
The clearest thing to admit up front: VoiceInk's Transcription History and "Retry Last Transcription" flow — storing the original text, the enhanced text, metadata, and the audio itself, with in-app playback and a global retry shortcut — is the best failure-recovery pattern found anywhere in this research, competitors included. It's a real bar to clear, and OpenWispr hasn't cleared it yet.
VoiceInk's "Retry Last Transcription," bound to a global shortcut, reruns the last recording with the current mode and model — deliberately designed for trying a different model on the same audio, not just re-sending the same request. Pair that with "Paste Last Transcription" and "Paste Last Enhanced Transcription" shortcuts for when the *insert* failed rather than the transcription, and VoiceInk covers failure modes most competitors, including OpenWispr today, don't yet handle.
Its hotkey ergonomics are also worth naming: single-modifier triggers (Right/Left Option, Cmd, Ctrl, Fn) with three activation styles — Toggle, Push-to-Talk, and a Hybrid mode where a short press toggles and a 0.5s+ hold becomes push-to-talk. That flexibility isn't matched in OpenWispr's current hotkey handling.
VoiceInk is Apple Silicon only — no Intel Mac support, no Windows, no Android at all. OpenWispr covers Android today and runs on both Apple Silicon and Intel Macs. If you're not on an M-series Mac, or you need a phone option, VoiceInk isn't in the running.
VoiceInk's packaged app also isn't free long-term: it's a one-time purchase of $29–69 to get auto-updates and support (building from source yourself is free, but that's a real barrier for most users). OpenWispr's MIT license is also more permissive than GPL-3.0 — no copyleft obligation if you build on top of it. Even VoiceInk's own open bugs are useful context: issue #830 (enhancement failures silently discarded) is exactly the failure class OpenWispr's own code review flagged as something to never do — emit nothing, silently.
Open source and entirely on your device. Android from Google Play, macOS from GitHub — no account, no sign-up, just a download.