Open AnyBrief
Install the signed and notarized build on a Mac running macOS 14 or newer.
whisper.cpp inside a Mac app
Use Whisper on Mac without command-line work
Choose the model and language in AnyBrief, then import a recording or capture a meeting. Local transcription uses the same processing workflow as recorded meetings.
Free beta · macOS 14+ · Apple Silicon and Intel
First run in three steps
No command line required.
Select Whisper
Choose a Whisper model and language in Recognition settings, enable Metal acceleration if needed, and download missing models.
Add a file or meeting
Import audio or video, or start recording. Once models are present, speech recognition runs locally.
What AnyBrief adds around whisper.cpp
An app workflow around the local model rather than a one-off transcript command.
Models and diagnostics
Select a Whisper model and language, and enable Metal acceleration when needed. Readiness checks cover tools and local model files; internet is needed for initial model downloads.
Fewer false phrases during silence
Required Silero VAD removes silent regions before recognition to reduce the risk of invented phrases.
Speaker separation
When enabled, Whisper word timestamps are aligned with local voice analysis. This separates turns but does not identify people.
Recording, Whisper and summaries together
- System audio and microphone capture
- Audio and video import
- Local transcript with speaker turns
- Cleanup, summaries and rule-based exports
Whisper questions
Does Whisper work offline?
Recognition runs locally. Internet is needed to download a missing model and for a remote LLM only if you choose one.
Does it run on Intel Macs?
AnyBrief supports Apple Silicon and Intel Macs. Speed depends on the computer and selected Whisper model.
Can I use another recognition engine?
Yes. AnyBrief also supports FluidAudio STT as a separate local transcription provider.
Use Whisper through a regular Mac interface
Set up the model once and process meetings and files.