OpenSuperWhisper is a macOS dictation app that provides real-time audio transcription using the Whisper model.
The tool solves the problem of converting spoken audio to text on macOS by offering a lightweight, locally-run transcription engine. It records audio through the system microphone and processes it using either the Whisper or Parakeet transcription models, which can be downloaded directly within the application. The approach emphasizes accessibility through global keyboard shortcuts, mouse button triggers, and hold-to-record modes that integrate seamlessly into the user's workflow.
The tool suits macOS users with Apple Silicon machines who want offline transcription without cloud dependencies. It works well for anyone needing quick voice-to-text conversion across applications, whether through keyboard shortcuts, mouse buttons, or by dragging audio files into the app for batch processing. The project supports multiple languages with auto-detection and includes specialized autocorrect for Asian languages. Users should be aware that the application currently requires Apple Silicon architecture and that Intel macOS compatibility remains on the contribution roadmap.
Development activity shows active engagement with user requests and feature expansion. The project maintains a public contribution roadmap that tracks both completed work and planned additions including streaming transcription, custom dictionary support, and agent mode functionality. Build infrastructure is documented through automated CI workflows, and the codebase uses git submodules to manage dependencies, indicating structured dependency management. The maintainers actively solicit contributions and provide clear guidance for local builds and troubleshooting.
