Skip to main content

Documentation Index

Fetch the complete documentation index at: https://help.heymumble.com/llms.txt

Use this file to discover all available pages before exploring further.

Mumble gives you two ways to process your voice: Cloud Mode and Local Mode. Both are full-featured. You choose per feature, based on what each conversation needs.

​
What’s the difference

Cloud Mode runs on our servers. You get the fastest processing, the highest accuracy, and support for 40+ languages. Audio is encrypted in transit and at rest, and is never stored after processing. Local Mode runs entirely on your Mac. Nothing leaves your device. Recording, transcription, AI Notes, and quick format templates all happen on-device. Works offline.

​
What runs locally

Local Mode covers Mumble’s core capture and notes:
FeatureCloudLocal
Meetings
(record + transcript + AI summary)		✓✓
Voice Notes✓✓
Dictation✓✓
AI Notes✓✓
Quick Format templates✓✓
Customized templates✓
Voice Skills✓
AI Chat✓
For Meetings, Voice Notes, and Dictation, Local Mode handles the full pipeline on your Mac, from audio capture to structured notes. Customized templates, Voice Skills, and AI Chat are Cloud only today. Meetings is the clearest example: recording, transcript, and AI summary all run on-device. You can take a meeting on a plane or in a secure room with no internet, and nothing about it (audio, transcript, or summary) ever leaves your Mac.

​
How to choose

You don’t have to pick one mode for everything. Each feature has its own toggle in Settings > Processing Mode. A simple way to think about it:
  • Cloud for everyday meetings and notes where you want the fastest results and broadest language coverage.
  • Local for sensitive conversations, regulated work, or anytime you want a clear privacy line. Also useful when you’re offline.
For a deeper look at what runs on-device and how to set it up, see About Local Mode.

​
Requirements for Local Mode

  • Apple Silicon Mac (M-series)
  • macOS 15 or later
  • 24GB unified memory or more recommended
  • One-time model download (about a few GB, done during onboarding or from Settings > Processing Mode)
A note on language: the local dictation model works best when you set a single language. Selecting multiple may reduce accuracy. Cloud Mode handles 40+ languages without this limitation.

​
Set it per feature

Open Settings > Processing Mode. Each feature has its own Cloud and Local toggle. Switch anytime.
Cloud and Local Mode