I built a language app when it first became viable with GPT and also went the avatar as UI route. It presents a unique set of challenges nd constraints, but I spent the most time just trying to get the mouth to sync with the audio. Fun experience for sure. Regarding learning languages, I have stopped building and relying on apps, as I spend too much time mucking with the app and not enough time on the language. The highest potency practice I have found is transcribing podcasts. It’s a major headache, but it really pushes you forward regarding listening, writing, and spelling.