Marvinngg

macOS native ambient voice input & meeting transcription — Apple SpeechAnalyzer + Vision OCR + FluidAudio speaker diarization, fully on-device.

19
3
100% credibility
Found Mar 20, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Swift
AI Summary

A native macOS app for on-device voice dictation, screen-aware transcription, and meeting recording with speaker identification and optional AI polishing.

How It Works

1
📱 Discover the voice helper

You find a free Mac app that turns your speech into perfect text for notes and meetings.

2
🔧 Set it up easily

Run a simple command to install it in your Applications folder and grant microphone and screen access.

3
It starts automatically

The app lives quietly in your menu bar, ready whenever you need it.

4
🎤 Speak with a key press

Press right Option key, talk naturally, press again—your words appear as typed text in any app.

5
🎙️ Record full meetings

Click the menu icon to start a meeting; a panel shows live transcripts with who spoke when.

6
Smart screen awareness

It glances at your screen for context, fixing names and terms automatically.

📝 Flawless notes ready

Export clean Markdown files of your meetings, improving each time you use it.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ambient-voice?

Ambient-voice delivers macOS-native ambient voice input and meeting transcription, fully on-device. Hit Right Option to dictate and auto-paste text into any app, with screen OCR feeding context for sharper recognition of code terms or proper nouns. For meetings, it runs continuous transcription via Apple SpeechAnalyzer, adds speaker diarization with FluidAudio, and exports Markdown—built in Swift for Apple Silicon on macOS 26.

Why is it gaining traction?

Zero-cloud privacy trumps web dictation tools, while screen context slashes errors in technical ambient voice recognition. Optional ollama polish refines spoken text to written form, and data collection enables self-improving distillation loops. Menu bar access and hotkey simplicity hook power users over clunky alternatives.

Who should use this?

MacOS devs dictating into native code editors or GitHub desktop clients during macOS GitHub actions workflows. Consultants or healthcare pros needing ambient voice transcription for notes, like ambient voice NHS meetings. Remote teams handling multi-speaker calls with quick Markdown exports.

Verdict

Grab it for macOS 26 ambient voice tech if Tahoe-ready—install via make, grant permissions, done. 19 stars and 1.0% credibility mean it's raw; solid ambient voice documentation helps, but expect tweaks. Promising starter for on-device dictation.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.