Muesli is a native macOS application for voice dictation, meeting transcription, and AI-assisted note-taking. It converts speech into text using local recognition models, with optional hosted providers. Users can dictate directly into application text fields through configurable keyboard shortcuts. Meeting recording captures microphone and system audio, with speaker labels and optional live captions. Transcripts can be transformed into structured notes, summaries, and searchable meeting records. An Ask Meetings interface answers questions using saved transcripts and provides references to relevant passages. Custom dictionaries improve recognition of names and technical terminology. The application also supports calendar integration, imported recordings, and Siri Shortcuts. iCloud synchronization transfers text and metadata between devices while keeping recorded audio local. Muesli requires an Apple Silicon Mac.
Features
- System-wide voice dictation with configurable hotkeys, hands-free recording, automatic text insertion, and filler-word cleanup
- On-device speech recognition through Whisper, Parakeet, Nemotron, Qwen3 ASR, and other supported models
- Meeting recording with microphone and system audio capture, speaker labels, live captions, and imported audio files
- AI-generated meeting summaries with customizable templates, Markdown or PDF exporting, and conversational transcript searching
- Custom vocabulary management with phrase replacements, JSON dictionary importing, and failed dictation recovery
- Calendar integration, Siri Shortcuts automation, optional hosted AI providers, and private iCloud synchronization of text and notes