Show HN: Privacy-First Voice-to-Text for macOS
6 hours ago
- #privacy-focused
- #speech-recognition
- #open-source
- High-quality speech recognition using WhisperKit with multiple model sizes for different accuracy/speed trade-offs.
- Supports multiple languages with auto-detection and features real-time waveform visualization during recording.
- Local LLM processing for grammar correction and text improvement with models like Gemma, Llama, Qwen, and Mistral.
- Custom prompts for grammar fixing, email formatting, language translation, and custom text processing workflows.
- 100% local processing ensures voice data never leaves the device with no cloud services or data collection.
- Open source and runs in a secure sandboxed environment for transparency and security.
- Features include global hotkey support, auto-copy to clipboard, auto-paste, auto-enter, and menu bar integration.
- Beautiful dark-themed interface with real-time recording visualization and comprehensive onboarding guide.
- System requirements include macOS 14.0 or later, 20GB free disk space, and necessary permissions.
- Easy installation via whisperclip.com or building from source with detailed setup and usage instructions.
- Privacy-focused design with local processing only, no network requests, no analytics, and encrypted storage.
- Open for contributions with guidelines for forking, feature branches, commits, and pull requests.
- Licensed under MIT License allowing commercial use, modification, distribution, and private use with attribution.