access_ai

AccessAI: Universal Web Accessibility Suite

AccessAI is an all-in-one Chrome extension designed to dismantle digital barriers. From AI-powered text simplification and smart voice-to-form dictation to reading modes and dyslexia-specific typography, AccessAI empowers users of all abilities to navigate the modern web with confidence.

---

🚀 Key Features

🧠 AI & Cognitive Support

• AI Text Simplification: Leverages Gemma 3 (via Hugging Face) to instantly simplify complex jargon into easy-to-read language.
• Support Levels: Choose from “Simple”, “Standard”, or “Detailed” based on your needs.
• OpenDyslexic Support: One-click toggle for specialized typography designed to improve reading speed and accuracy for dyslexic users.

🔊 Advanced Audio & Voice

• Universal Form Assistant: Intelligent voice-input for any text field, select dropdown, or radio group.
• Local Transcription: Uses faster-whisper running locally on your private proxy server—no external STT APIs or quotas required.
• NLP Email Parsing: Intelligent dictation formatting—automatically converts spoken “at” into “@” symbols and handles digit sequences.
• TTS Queue Manager: High-precision text-to-speech with audio confirmations for every UI action.

👁️ Visual & Layout Customization

• Color Filters: Built-in overlays for Protanopia, Deuteranopia, and Tritanopia.
• Reader Mode: Distraction-free reading by stripping away ads, popups, and non-essential layout elements.
• Dynamic Font Scaling: Real-time page-wide font scaling that respects site architecture.

---

📁 Project Structure

├── manifest.json         # Extension configuration
├── assets/               # Brand icons and visual assets
├── background.js         # Service worker for TTS, logic, and context menus
├── content.js            # Core UI injection, simplifies, and page logic
├── popup.html            # Primary settings dashboard
├── sidePanel.js          # Logic for side panel analysis and image OCR
├── opendyslexic-0.92/    # Integrated accessibility font files
└── proxy/                # Python backend (Hugging Face Space)
    ├── app.py            # Flask API for STT and LLM logic
    └── gemini_service.py # Integration with Gemini 2.0 and Gemma 3

---

🛠️ Getting Started

To download the dependencies and to run this locally. This project consists of two parts: The Chrome Extension (Client) and The Private Proxy (Server running on Hugging Face).

Prerequisites

1. A Google Gemini API Key.
2. A Hugging Face Account and Access Token (HF_TOKEN).

Setup the Proxy

1. Create a new “Docker” Space on Hugging Face.
2. Upload the contents of the /proxy folder.
3. Add your GEMINI_API_KEY, HF_TOKEN, and PROXY_API_KEY (custom password) to the Space Secrets.

Install the Extension

1. Update PROXY_URL and PROXY_API_KEY in content.js and sidePanel.js.
2. Open chrome://extensions/ in Chrome.
3. Enable “Developer mode” and click “Load unpacked”.
4. Select the project folder.

---

🔒 Security & Privacy

• EndPoint Protection: All incoming proxy requests are authenticated via a custom X-AccessAI-Key header.
• Local Audio Privacy: Transcription is performed locally on your private server using Whisper—no third-party cloud audio recording.

📜 License

Distributable under the MIT License.

This site is open source. Improve this page.