DooDooLamb News
Show HN: Dictata – Local Whisper dictation with LLM cleanup
Brief published August 16, 2026 · Original source published August 15, 2026
Original reporting by AntoineC_ at github.com.
Automated brief. Verify important details at the original source.
What happened
Dictata is a newly released open-source tool for Windows that provides system-wide voice dictation running entirely on-device. Written in native Rust, it uses whisper.cpp for transcription and an LLM for text cleanup. The workflow is hotkey-triggered: press once to start recording, speak, press again, and the transcribed text is pasted into whatever application is active. The project is hosted publicly on GitHub.
Why it matters
Fully local dictation avoids sending audio or text to external APIs, which matters for developers handling sensitive data or working in air-gapped environments. Combining whisper.cpp with an LLM cleanup step addresses a common complaint about raw transcription quality without requiring a cloud subscription.
What to watch
Adoption will depend on which whisper.cpp model sizes are supported and what hardware is required to run both the transcription and LLM cleanup at acceptable latency. Windows-only scope limits reach, so watch whether Linux or macOS support is planned. Community traction is minimal so far, with only two points and no comments at time of publication.