Inspiration
Pronounciation Analysis is something most people cannot do without a skilled teacher. Duolingo or Google can't teach you how to be fluent in a language, only talking with a real person or artificial intelligence capable of analyzing your words can make you fluent.
What it does
Takes a diagnostic test. Generates a pathway for learning. Creates exercises and tracks your goals. Have a 1-on-1 conversation where AI would analyza your pronounciation and word selection.
How we built it
This is an MVP utilizing Streamlit for the front-end; it utilizes Gemma 3 as the main AI backend. It uses Montreal Forced Aligner with phenome libraries of languages to analyze pronounciation. Vosk for speech to text and Edge TTS for text to speech.
Challenges we ran into
Locating and tweaking the right AI model for the job.
Built With
- gemma
- streamlit
- vosk
Log in or sign up for Devpost to join the conversation.