by jpcom on 3/24/25, 1:38 PM with 5 comments
by jpcom on 3/24/25, 1:42 PM
Aiko uses xAI’s Grok (latest model) to craft natural Japanese replies and AWS Polly (Mizuki voice) to speak them. Pick a topic like "Greetings" or "Ordering Food," hit "Click to Record," and talk to her in Japanese—she listens via Web Speech API, responds, and plays it back. It’s real-time, with a mic level visualizer and auto-scrolling chat, all in client-side JS (React + Tailwind).
This ties into my Japanese Complete project (data-driven, 90% real-world Japanese coverage). Sold a $450 lifetime plan this morning, so I figured it’s HN time. Next up: server-side polish, auth, and more topics. Would love your take—especially on UX or scaling ideas!
by mreichhoff on 3/24/25, 3:54 PM