VoiceLink Community is a maintained fork of the original VoiceInk project focused on a fully unlocked, offline-first macOS transcription workflow. It keeps the privacy-first core, bundles local models, and adds community-friendly defaults for features like Power Mode, Text-to-Speech, and rapid keyboard control.
Highlights of this fork:
- 💸 Fully unlocked – no trials, license prompts, or gated features.
- 📦 Models included – Whisper (multiple sizes) and Parakeet ship in the app, ready to use offline.
- 🔧 Hackable by default – a friendlier contributing policy and cleaner onboarding for builders.
- 🎙️ Offline transcription – Whisper, Parakeet, and Apple Speech models are bundled and ready on first launch.
- 🔊 Text-to-speech studio – Create narration with OpenAI, ElevenLabs, Google Cloud, or local system voices, complete with previews, batch queueing, translation, and article import tools.
- 🔒 Privacy first – audio and transcripts stay local unless you explicitly export them.
- ⚡ Power Mode – detect the active app/URL and auto-apply prompts, models, and paste rules.
- 🎯 Global shortcuts – flexible hotkeys, push-to-talk, and middle-click control.
- 📝 Custom vocabulary – dictionaries, word replacements, and CSV import/export.
- 💬 Optional enhancements – local formatting works out of the box; Ollama hooks stay available for power users.
- Releases – check the repository releases tab for notarized builds of the community edition.
- Homebrew (optional) – once a tap is available you’ll be able to
brew install --cask voiceink-community. - From source – follow BUILDING.md to compile the app with Xcode. Run
./scripts/download-models.shbeforehand to drop the default Whisper binaries into the bundle if you haven’t downloaded them yet.
- macOS 14.0 or later
- Building from Source - Detailed instructions for building the project
- Documentation Index - Feature guides and workflows for day-to-day use
- Power Mode Guide - Context-aware automation by app or URL
- Text-to-Speech Workspace Guide - Narration, batch generation, and export
- AI Enhancement Guide - Providers, prompts, and context settings
- Keyboard Shortcuts Guide - Hotkeys, push-to-talk, and actions
- Dictionary Guide - Quick rules, replacements, and vocabulary
- Model Management Guide - Local, cloud, and custom models
- Data Management Guide - History, export, and cleanup
- Contributing Guidelines - How to contribute to VoiceLink Community
- Code of Conduct - Our community standards
- Rectifications & Improvements - Security and performance fixes applied to the community edition
- Changelog - Release-by-release changes
Recent stability, security, and performance improvements are documented in
VOICELINK_COMMUNITY_REMEDIATIONS.md. Highlights include:
- ✅ HTTPS validation for custom AI provider verification.
- ✅ Non-blocking audio file handling for cloud transcription.
- ✅ Reduced main-thread hopping in
@MainActorclasses. - ✅ Removal of forced
UserDefaults.synchronize()in hot paths. - ✅ Streamed audio preprocessing and transcription uploads to reduce memory.
- ✅ Disk-cached recent TTS history audio with size limits and cleanup.
- Refactoring: Centralized settings management in
AppSettings. - Cleanup: Relocated
AuthorizationHeaderand consolidated utilities.
- Performance: streamed audio preprocessing and transcription uploads.
- Memory: capped OCR/browser context and stored AI request payloads.
- Storage: cached recent TTS history audio on disk with size limits.
Pull requests are welcome without prior approval. Read the lightweight CONTRIBUTING.md for tips, spin up a branch, and open a PR when ready. If you want early feedback, drafts and GitHub Discussions are perfect places to start the conversation.
This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.
If you run into trouble:
- Search existing GitHub issues and discussions.
- Open a new issue with logs, screenshots, or steps to reproduce.
- Join the Discord (linked inside the app) for quick questions or pairing sessions.
- whisper.cpp - High-performance inference of OpenAI's Whisper model
- FluidAudio - Used for Parakeet model implementation
- Sparkle - Keeping VoiceLink Community up to date
- KeyboardShortcuts - User-customizable keyboard shortcuts
- LaunchAtLogin - Launch at login functionality
- MediaRemoteAdapter - Media playback control during recording
- Zip - File compression and decompression utilities
- SelectedTextKit - A modern macOS library for getting selected text
- Swift Atomics - Low-level atomic operations for thread-safe concurrent programming
Maintained with ❤️ by the VoiceLink Community