I built a live captioning system for local technical talks back in 2024. Back then it requires putting multiple models together so it works well enough with Thai language + tech jargons. https://www.youtube.com/watch?v=rXMS212fZSA
The models have improved much since then, so curious to see how it would look like if rebuilt from scratch…
2024 version: https://github.com/dtinth/live-speech
commit 835aa2e
Here! Fresh orphan branch
v2 is checked out and empty. Where do you want to start: the transcription model, the audio capture side, or the viewer?Newly built live captioner validated at venue.
- Qwen Audio 3.1 ASR Flash Streaming's ASR result is not usable on its own.
- The clutch was building a REST API for Claude to read and refine the transcript. It also occasionally pulls the livestream video frame for better fixes.
-…