https://github.com/phantom-ear/phantom-ear
PhantomEar is a privacy-first, local-only AI meeting assistant that runs invisibly on your machine. No bots, No cloud recording, No admin permissions.
https://github.com/phantom-ear/phantom-ear
ai-agents privacy rag-chatbot stt
Last synced: 4 months ago
JSON representation
PhantomEar is a privacy-first, local-only AI meeting assistant that runs invisibly on your machine. No bots, No cloud recording, No admin permissions.
- Host: GitHub
- URL: https://github.com/phantom-ear/phantom-ear
- Owner: Phantom-Ear
- Created: 2026-02-06T18:10:12.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2026-02-21T22:49:51.000Z (4 months ago)
- Last Synced: 2026-02-23T09:36:52.702Z (4 months ago)
- Topics: ai-agents, privacy, rag-chatbot, stt
- Language: Rust
- Homepage: https://phomy.mosaabyassir.xyz
- Size: 1.5 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 👻 PhantomEar (Now Fomy.io)
### **Always Listening. Never Seen.**
*The privacy-first AI meeting assistant that lives entirely on your machine.*
**✨ Visit our new home: [fomy.io](https://fomy.io) ✨**
[](https://fomy.io)
[](https://github.com/Phantom-Ear/phantom-ear/releases/latest)
[](https://github.com/Phantom-Ear/phantom-ear/releases)
[](LICENSE)
[](https://github.com/Phantom-Ear/phantom-ear/releases)
[](https://github.com/Phantom-Ear/phantom-ear/actions)
[](https://www.rust-lang.org/)
[](https://tauri.app/)
[](https://svelte.dev/)
[**Download**](#-download) • [**Features**](#-features) • [**Phomy AI**](#-meet-phomy) • [**How It Works**](#-how-it-works) • [**Tech Stack**](#-architecture)
---
## 🎯 The Problem with Modern Meeting AI
Traditional meeting assistants **expose your presence**:
- 🤖 **Bots join the call:** Everyone knows you're recording.
- ☁️ **Cloud uploads:** Your private conversations leave your secure environment.
- 🔒 **Admin privileges:** IT blocks the installation of invasive tools.
- 💰 **Expensive subscriptions:** Monthly costs rack up for simple transcription features.
## ✨ The PhantomEar Solution
PhantomEar (fomy) is fundamentally different. It is an ambient intelligence tool built for privacy purists.
### 🔒 **100% Local Processing**
Audio never leaves your machine. Transcription runs locally using optimized Whisper models. Your meetings, your data, your control.
### 👻 **Zero Meeting Footprint**
No jarring bots invading your Zoom calls. No "Recording in progress" banners. No participant prompts. Completely invisible.
### 🧠 **Intelligent Recall**
Ask questions about any meeting. Get instant algorithmic summaries. Search across all your conversations effortlessly with bleeding-edge semantic search.
### 🏢 **Enterprise Ready out of the Box**
User-level installation. Runs smoothly without admin rights or IT tickets. Functions perfectly on restricted corporate laptops.
---
## 📥 Download
Ready to take back your meeting privacy?
### 🚀 **[Download Latest Release](https://github.com/Phantom-Ear/phantom-ear/releases/latest)**
*(Or visit [fomy.io](https://fomy.io) for more information)*
| Platform | Architecture | Binary |
|----------|-------------|----------|
| **macOS** | Apple Silicon (M1/M2/M3) | [`.dmg`](https://github.com/Phantom-Ear/phantom-ear/releases/latest) |
| **Windows** | x64 | [`.exe`](https://github.com/Phantom-Ear/phantom-ear/releases/latest) / [`.msi`](https://github.com/Phantom-Ear/phantom-ear/releases/latest) |
🛠 Troubleshooting Installation
### macOS
Since PhantomEar is open-source and not yet notarized with Apple, you may see an "App is damaged" warning. Fix it instantly with:
```bash
xattr -cr /Applications/PhantomEar.app
```
### Windows
- If `.msi` is blocked by strict corporate policy, use the portable `.exe` installer.
- If you see an `MSVCP140.dll` error on a fresh Windows system, install the [Visual C++ Runtime](https://aka.ms/vs/17/release/vc_redist.x64.exe).
---
## 🎬 Core Features
Real-Time Transcription
Live, hyper-accurate speech-to-text powered by local Whisper.
AI Q&A
Interact naturally and ask hyper-specific questions about any past meeting.
Auto Summaries
Distill hour-long meetings into key points, decisions & action items.
Semantic Search
Find abstract concepts or exact phrases across all your stored meetings.
Persistent Memory
Robust SQLite backed with full-text fuzzy search.
Air-Gapped Privacy
Everything stays on your device. Period.
### 👻 Meet Phomy: Your Intelligent Meeting Co-Pilot
Phomy is your dedicated AI meeting assistant. It uses advanced RAG (Retrieval-Augmented Generation) to understand your intent:
- **Time Windows:** *"What did we discuss in the last 15 minutes?"*
- **Action Items:** *"List all the tasks assigned to me during this sync."*
- **Global Search:** *"When did we last talk about the Q3 marketing budget across all meetings?"*
- **Web Fallback:** If you ask a question outside the scope of your transcripts (e.g. *"Who won the world cup in the year we discussed?"*), Phomy seamlessly falls back to a live, configurable Web Search that you can cancel at any time.
### Feature Highlights
- 🎙️ **Multi-Model ASR:** Choose between varying Whisper variants (*tiny → large*) or Parakeet CTC models based on your hardware specs.
- 🤖 **LLM Flexibility:** Plug in an OpenAI API key or use **Ollama** (for a fully air-gapped, 100% local operation).
- ⏸️ **Pause/Resume:** Pause transcription on-the-fly without breaking the recording session.
- 📝 **Transcript Editing:** Modify generated segments or fix rare transcription errors.
- 🗣️ **Speaker Labels:** Manually tag speakers to dialogue segments for better context tracking.
- 📊 **Visual Timeline:** Navigate long recordings effortlessly with an interactive timeline.
- 🔍 **Quick Global Search:** Hit `Cmd+K` from anywhere to search across every recorded meeting.
- 🖥️ **System Tray Integration:** Run headlessly in the background. Start/stop directly from the menu bar.
- 📋 **Seamless Export:** One-click copy interactions as Markdown or plain text to share.
---
## 🔬 How It Works Architecture
```text
┌─────────────────────────────────────────────────────────────────────────┐
│ YOUR MACHINE │
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌──────────┐ │
│ │ Microphone │───▶│ Whisper │───▶│ SQLite │───▶│ UI │ │
│ │ (cpal) │ │ (local) │ │ + FTS5 │ │ (Svelte) │ │
│ └─────────────┘ └─────────────┘ └─────────────┘ └──────────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌─────────────┐ ┌─────────────┐ │
│ │ Embeddings │ │ Semantic │ │
│ │ (BGE-small) │───▶│ Search │ │
│ └─────────────┘ └─────────────┘ │
│ │ │
└───────────────────────────────────────────────│─────────────────────────┘
│
┌─────────────────────┴─────────────────────┐
│ OPTIONAL: LLM Context │
│ (Only curated text, never raw audio) │
└─────────────────────┬─────────────────────┘
│
┌───────────┴───────────┐
▼ ▼
┌──────────┐ ┌──────────┐
│ Ollama │ │ OpenAI │
│ (local) │ │ (API) │
└──────────┘ └──────────┘
```
### Data Privacy Contract
| Data Type | Location | Sent to Cloud? |
|-----------|----------|----------------|
| 🎤 Raw Audio | Ephemeral RAM only (never stored) | ❌ **Never** |
| 📝 Transcripts | Encrypted Local SQLite | ❌ **Never** |
| 🧠 Embeddings | Encrypted Local SQLite | ❌ **Never** |
| 💬 LLM Context | Snippet text | ⚠️ **Optional*** |
*\*Only if explicitly using the OpenAI API integration. Switch to Ollama for a 100% local, air-gapped data loop.*
---
## 🏗️ Development & Contribution
### Tech Stack Breakdown
Rust
Tauri 2.0
Svelte 5
TailwindCSS
### Getting Started Locally
```bash
# 1. Clone the project
git clone https://github.com/Phantom-Ear/phantom-ear.git
cd phantom-ear
# 2. Install Node dependencies
npm install
# 3. Spin up the Tauri Dev Server
npm run tauri dev
# 4. Build for Production Release
npm run tauri build
```
*Note: You must have Node v18+, Rust 1.75+, and CMake (for whisper.cpp) installed.*
## 🤝 Contributing
PhantomEar is built by the community. Check out our ongoing Roadmap to see what we're working on next!
1. Fork the repo.
2. Create your feature branch (`git checkout -b feature/amazing-feature`).
3. Commit your changes (`git commit -m "Add amazing feature"`).
4. Push to the branch (`git push origin feature/amazing-feature`).
5. Open a Pull Request.
---
## 📄 License & Legal
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
### Built with 🖤 for privacy advocates at [fomy.io](https://fomy.io)
**[⬆ Back to Top](#-phantomear-now-fomyio)**
PhantomEar is not affiliated with any meeting platform (Zoom, Teams, Google Meet, etc). Please use responsibly and ensure you comply with all local and applicable laws regarding audio recording.