I built a real-time pipeline that reads game subtitles and converts them into dynamic voice acting (OCR → TTS → RVC) [P]

About this article

I've been experimenting with real-time pipelines that combine OCR + TTS + voice conversion, and I ended up building a desktop app that can "voice" game subtitles dynamically. The idea is simple: - Capture subtitles from screen (OCR) - Convert them into speech (TTS) - Transform the voice per character (RVC) But the hard parts were: - Avoiding repeated subtitle spam (similarity filtering) - Keeping latency low (~0.3s) - Handling multiple characters with different voice models without reloading ...