ai-pronunciation-trainer

Why we included this project

This tool attacks the classic problem of not being able to hear your own accent. You read a sentence out loud, and it uses Whisper to transcribe what you said, then scores how closely your production matches the target pronunciation. That active speaking loop makes it more useful than passive listening apps, and the setup is straightforward: pip install a couple things, have ffmpeg ready, and run webApp.py, or hit the hosted version if you don't want to fiddle with code. It started as one person's hobby project, so the code has rough edges and the feature set is lean, but for self-study or a light classroom drill it holds up. Adding a new language mostly means Whisper needs to support it, since the rest is pluggable.