Overview / Description
MumbleFlow is an AI speech-to-text app for macOS that transcribes your voice entirely on-device, so recordings never leave your Mac. It is built for Mac users who want fast dictation without a cloud subscription or the privacy trade-off of uploading audio to a remote server. Transcription runs on whisper.cpp with Metal or CUDA acceleration, and a global Fn hotkey lets you start dictating in any application, with text inserted directly at your cursor. A local LLM powers a "Smart Cleanup" pass that adds punctuation, fixes grammar, and formats the output, while a real-time waveform shows recording status as you speak. Because all processing is local, MumbleFlow discards the audio immediately rather than storing it, which is the core reason to pick it over cloud dictation tools. It currently requires macOS 14 or later; the vendor lists Windows and Linux support as coming soon, so non-Mac users cannot use it yet. It is sold as a one-time $5 purchase with no subscription, positioned against cloud services that charge monthly or yearly. As a Mac-first local dictation tool, it trades broad platform support for privacy and a flat price.
Used For
On-device voice dictation and transcription on macOS across any application, with privacy-first local processing
Pricing
Pros & Cons
Pros
- On-device transcription via whisper.cpp with Metal/CUDA acceleration — no cloud upload
- Global Fn hotkey dictation that works in any app, inserting text at the cursor
- Local-LLM "Smart Cleanup" for punctuation, grammar, and formatting
- One-time $5 purchase with no subscription
- Audio discarded immediately rather than stored, for privacy
Cons
- macOS 14+ only; Windows and Linux support is not yet available
- One-time price means transcription quality depends on the bundled local models
- No cloud sync or multi-device account, given the local-only design
- Newer indie app, so integration breadth is limited
Questions & Answers
Alternatives
Whisper, MacWhisper, Superwhisper, Wispr Flow