Back to projects

Systems

VoiceKey

Offline, privacy-first voice-to-keyboard for Linux and Windows

Pythonfaster-whisperPortAudio / ALSAffmpegCLIMkDocs
Repo

Role

Creator & Maintainer

Duration

2026

Date

2026

Type

Personal Tool

By the Numbers

259

Commits

MkDocs

Docs Site

Yes

Release Pipeline

Overview

VoiceKey types what you say, entirely offline, using faster-whisper. It runs as a daemon or in the foreground, ships a CLI with status, devices, and config commands, and offers an explicit opt-in hybrid mode with an OpenAI-compatible cloud fallback. Local stays the default. The installer supports skip flags and model profiles (tiny, base, small), and the project is packaged the way real tools are: CI plus a release workflow, a MkDocs documentation site, and architecture, requirements, security, and changelog documents.

Tech Stack

Pythonfaster-whisperPortAudio / ALSAffmpegCLIMkDocs

Inside the Project

What it does, how it works, and what it ships.

lock

Local First, Cloud Optional

Local faster-whisper is the primary engine; the cloud fallback is an explicit opt-in through an OpenAI-compatible endpoint. Nothing leaves the machine unless you turn it on.

build

Built Like a Product

pip-installable with a pyproject, a one-command installer with --skip-system-deps, --skip-model-download, --autostart, and --model-profile flags, plus daemon mode with a documented background contract.

  • voicekey start --daemon, status, devices, config
  • Model profiles: tiny / base / small
  • Monthly patch cadence with CHANGELOG
star

Engineering Hygiene

  • CI and release workflows on GitHub Actions
  • MkDocs documentation site
  • architecture.md, software_requirements.md, SECURITY.md, CONTRIBUTING.md

Outcomes

A daily-driver utility with an uncommon level of packaging discipline for a personal tool.

mic

Offline

Local-First Typing

terminal

CLI

Daemon + Commands

commit

259

Commits

Interested in this project?

Reach out for source code walkthroughs or collaboration.

RepoView All ProjectsContact