AHGlasses
Wearable real-time OCR and speech-to-text subtitle system.
Overview
AHGlasses is a wearable system that provides real‑time OCR and speech‑to‑text subtitles. The current prototype uses a Pi Zero W and camera module mounted on a pair of glasses, with a laptop handling OCR, speech recognition, and subtitle rendering. The project explores lightweight wearable capture and real‑time processing. Future updates may include improved OCR accuracy, better voice handling, and a more refined hardware design.
Architecture
AHGlasses uses a split design. The Pi Zero W captures camera input and streams it to a laptop, which performs OCR, speech recognition, and subtitle generation. This setup keeps the wearable light while offloading the heavier processing to the laptop.
System Architecture Diagram
High‑level view of camera, microphone, server, and subtitle renderer.
Processing Pipeline
The pipeline includes camera capture, microphone input, OCR preprocessing, Vosk speech recognition, and HTML overlay rendering. The system produces live subtitles in a browser window.
Data Flow / Pipeline Diagram
Data flow from camera and microphone to the live subtitle overlay.
User Interface
The browser‑based UI displays OCR and speech recognition results as live subtitles. It is simple, responsive, and works in varied lighting and audio environments.
Subtitle Display
HTML overlay showing real‑time OCR and speech‑to‑text output.
Development Notes
Development focused on stabilizing the capture pipeline, improving OCR preprocessing, and handling noisy audio. The current prototype works reliably for testing. Future plans include refining the OCR and voice models and exploring a more compact hardware setup.