Japan Cat
Learn Japanese while driving. Don't waste time.
24 followers
Learn Japanese while driving. Don't waste time.
24 followers
You can read some Japanese and know the words, but freeze the moment you have to speak. Japan Cat fixes that. Talk to an AI in Japanese and it replies naturally, then shows how to say it better, what you got wrong, and the nuance. Pick a partner for casual chat or structured tutoring. Review past conversations and retry. No real person, no embarrassment. Works hands-free in landscape and on CarPlay for commute practice.
Audio-only while driving is a real constraint and I like that the product commits to it rather than pretending it's a full study app.
The part I'd want to know is how recall works without hands. Most learning apps lean on the user tapping "I knew this / I didn't," and that signal is what drives the spacing. Driving takes that away entirely, so either the schedule is fixed, or it's inferring something from voice, or it just repeats on a timer. Which is it?
The other thing audio-only makes hard is pitch accent. 橋 and 箸 are the same string of sounds and different words, and learners who only ever hear things at speed tend to flatten both. Is that something the course addresses directly, or is it aimed more at getting to functional listening first?
@ark_y_k Hi Yuki-san I really appreciate your comments!
In this app, drivers don't really touch the screen at all. All you need to do is to keep the conversation ongoing. You can reply to JapanCat saying you 'don't know a word - can you please elaborate', or anything like that.
The app gradually remembers your preferences, keeps track of the mistakes you make, and can quiz you on them again later.
Regarding pitch accent, the audio-only mode helps address this because JapanCat usually speaks the entire sentence aloud rather than presenting isolated words. This gives the AI enough context to distinguish words such as hashi based on their meaning and pronunciation.
@pursuit_li Ah, so the "I don't know this word" turn is itself the input — that's a neater solution than I gave you credit for. Asking is something a driver can do without looking, and it doubles as the signal you'd otherwise have to collect with a button.
One thing I'd add as a native speaker, if it's useful: what learners lose first in audio-only is pitch. 雨 and 飴 are the same syllables with different shapes, and at conversational speed the two collapse into one word if nobody ever flags the difference. Learners often don't discover it until someone doesn't understand them.
Not necessarily something to teach explicitly — but if the TTS voice keeps it consistent, people absorb it without being told. Is it a synthesized voice or recorded audio?
@ark_y_k Hi Yuki-san. I understand where you come from as the pitch accent matters. I have a native Cantonese background and it functions similarly too. The app uses ElevenLabs’ Japanese TTS rather than the default browser voice, so the pronunciation and prosody are much more natural. Your comment is a good reminder that I should also check how consistently it handles pitch accent across different words and sentences!
@pursuit_li A Cantonese background is about the best possible starting point for this — you already treat tone as part of the word rather than as decoration, which is the exact thing most learners from non-tonal languages never quite internalize about Japanese.
And ElevenLabs over browser TTS is the right call for this specifically. The default voices tend to flatten accent entirely, so learners absorb a version of the word that doesn't exist. If the pitch is consistent in the audio, people pick it up without ever being taught it — which for a driving app is the only way it could work anyway.
Good luck with the launch. Nice to see a Japanese learning tool that picks one situation and builds properly for it instead of trying to be everything.