Voice routed to where it belongs.

blackrim-vox gives developers a local-first voice pipeline with Wispr Flow parity on macOS, Linux, and Windows. Hotkey-triggered dictation routes transcribed text directly to your editor, terminal, or any input field — no audio leaves your machine unless you explicitly configure a cloud ASR backend. Meeting diarization is on the roadmap (see ADR-0005). Privacy is structural, not a setting.

Apache 2.0macOS / Linux / WindowsLocal-first

Local-first processing

Audio stays on device. No cloud transcription required. BYOK when you want cloud ASR; local Whisper when you do not.

Cross-platform routing

macOS, Linux, and Windows. Hotkey-triggered dictation with configurable ASR backends and output targets.

Open source, Apache 2.0

No vendor lock-in. Inspect the pipeline, swap components, run it in air-gapped environments.