Local macOS meeting transcription and dictation app using Apple Silicon.
Agent-native local meeting transcription and dictation for macOS as an open source alternative.
Muesli is a native macOS app combining local dictation and meeting transcription powered by Apple Silicon. It suits professionals seeking privacy-first voice tools with optional hosted backends.
It captures microphone and system audio simultaneously, utilizing VAD-driven chunking, speaker diarization, and local or hosted models. You can dictate text at your cursor, rewrite text with Quill, and manage calendar events.
It suits macOS professionals handling meetings and dictation who prioritize local privacy and Apple Silicon performance. It does not suit Windows or Linux users since it is strictly a native macOS application.
Achieves low latency dictation via local models running on the Apple Neural Engine.
Captures your microphone and system audio simultaneously using a CoreAudio process tap.
Silero VAD detects natural speech boundaries to split audio at pauses without mid-sentence cuts.
Identifies individual remote speakers in system audio using CoreML diarization models.
Highlight text and speak an instruction to rewrite it or generate text at your cursor.
Free under the MIT licence. Optional hosted providers like OpenAI, Anthropic, or OpenRouter require your own paid API keys or subscriptions.
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
No, dictation and meeting transcription run locally on Apple Silicon by default unless you explicitly opt into hosted providers.
Yes, you can use your own OpenAI, Anthropic, or OpenRouter API keys for hosted transcription and summaries.
It integrates with your Mac calendar to detect upcoming meetings, display event notifications, and offer a Join and Transcribe action.
See Zekai first in Google