Tech: Next.js, pdf.js, Tesseract.js, Web Audio API, VAD, Tailwind
License: MIT (open source)
Why
Actor friends often rehearse alone and have to tap screens or read both parts. I wanted to see if a browser could listen to live lines and speak the next cue automatically.
What I Built
Load any script PDF (including scanned ones)
The browser listens while you speak
When you finish a line, it speaks the next one
No tapping, works offline after load
How
pdf.js + Tesseract.js for text extraction
Web Audio API + VAD for turn detection
Fuzzy matching so imperfect delivery still advances