# At-Table

**Bridging the gap between Deaf and Hearing conversations through local mesh networking and calm design.**

At-Table is an iOS accessibility application designed to enable seamless, simultaneous conversations between multiple deaf and hearing individuals. By leveraging a full mesh network, devices connect automatically without the need for Wi-Fi routers or cellular data, creating a local "table" of communication.

The app prioritizes a stress-free environment for deaf users, utilizing calming visuals to counteract the anxiety often associated with following rapid-fire hearing conversations.

## 🌟 Key Features

### Core Communication
* **Full Mesh Networking:** Devices enter a peer-to-peer full mesh network using Apple's Network.framework (NWConnection/NWBrowser) with AWDL support. No manual pairing or internet connection is required to chat.
* **Real-Time Transcription (Hearing Role):** Hearing users speak naturally. The app transcribes speech in real-time and automatically posts the text as a message to the group once the speaker pauses.
* **Text & Quick Replies (Deaf Role):** Deaf users can type messages or use one-tap "Quick Reply" chips (Yes, No, Hold on, Thanks) for rapid interaction.
* **Role-Based UI:** Distinct interfaces tailored for "Deaf/HoH" and "Hearing" users, optimized for their specific communication needs.

### Visual Design & Atmosphere
* **Calm ASMR Aesthetic:** The background features a slow-moving, blurring mesh of "Liquid Glass" orbs. This visual ASMR is designed to induce a calm mood, helping alleviate the cognitive load and stress deaf users experience during hearing conversations.
* **Personalized Auras:** Users select a "Neon Aura" color during onboarding. This color identifies them in the mesh and dynamically influences the background animation of their device.
* **Live Transcript Streaming:** Hearing users' speech appears as "partial" streaming text on receiving devices before becoming a finalized message, allowing for faster reading speeds.

### Technical Resilience
* **Ghost Filtering:** Advanced logic prevents "ghost" connections (stale peer signals) from clogging the network.
* **Smart Recovery:** Includes "Identity Thrashing" prevention and auto-reconnection logic to handle interruptions or app backgrounding seamlessly.
* **Hybrid Speech Recognition:** Utilizes server-based speech recognition for high accuracy when online, with a seamless fallback to on-device recognition when offline.

## 📱 How It Works

1.  **Onboarding:** Users enter a display name, select their role (Deaf/HoH or Hearing), and choose an identifying color.
2.  **Discovery:** The app automatically advertises and browses for nearby devices running At-Table.
3.  **Connection:** Devices negotiate a connection automatically using a leader/follower algorithm to prevent collision loops.
4.  **The Conversation:**
    * **Hearing users** simply hold their phone; the microphone listens for speech, visualizes it, and broadcasts it.
    * **Deaf users** read incoming bubbles and participate via text.


## 🔒 Privacy

* **Transient Data:** Messages and transcriptions are ephemeral. They exist only for the duration of the session and are not stored in any persistent database or uploaded to a cloud server (other than the temporary audio buffer sent to Apple for transcription if online).
* **Local Connectivity:** Chat data flows directly between devices over Wi-Fi/Bluetooth peer-to-peer protocols.