I've built a tool that uses AI for some of its translation (
https://nuenki.app). It's a browser extension that translates relevant sentences into your target language, and it uses Claude for text that isn't yet visible.
So I spent a decent amount of time in this area.
There are a lot of services that offer an "AI tutor", to the point that it causes difficulty for me marketing, because everyone has seen it 300 times and assume my product is the 301st (it's not). They're so numerous and similar that I can't actually think of many, because they all merge together.
I know Duolingo has one under a $20/month subscription, though I've not tried it.
On the whole, 99% of the services are just thin wrappers around the standard LLM APIs. If you just want text conversations, just ask a decent LLM (I can vouch for Claude, fwiw) to tutor you.
Pronunciation is a much more interesting topic. I've had the same idea - using a multimodal LLM to give precise feedback on accents. If I knew data science, I'd implement it myself! I heard from a friend who does data science that Duolingo has some papers on it, but I'm yet to see a service that actually offers it.
GPT-4o voice mode and Gemini might be able to do it if you ask, but other than that, well, I've kept an eye out and I'm yet to see one. Let me know if you find something.
Edit: Cambridge has something, but it's only for English.
https://cambridgeschool.online/product/cambridge-ai-speech-t...
