Nice, you support the big ones that many of LR’s community has been asking for.
I’ve been saying for a while now that LR needs to shift into high gear and deliver some major features, but it feels like we’re stuck in “maintenance mode.” At this point, next-gen language tools—many of which took inspiration from LR—are quickly surpassing it in features.
For LR’s long-term success, it might be in its best interest to sell to a larger company with the resources to develop it into the platform it deserves to be.
I mean no disrespect to LR’s dev team—I know it’s a very small group—but the current pace is holding the platform back instead of allowing it to grow.
Regarding FluentAI, how is the Japanese parsing handled? The website looks great and offers some excellent features, but parsing has always been a weak point with Asian languages—especially Japanese. Even LR often misses many compound words, as well as negative, passive, and other conjugated forms.
Basically I’ve made a lot of research, talked to a lot of Japanese learners. I’ve tested many libraries in JavaScript that attempt to do the Japanese parsing. But the conclusion I had with the Reddit Learn Japanese, for instance, is that the large language models are becoming the best at parsing Japanese.
So in the app, we are using a pre-fetching system that prefetches the tokenization for the Japanese, and hovering a word will take into consideration this tokenization.
The pros is that it’s usually more precise than the packages available, and as the large language models will become stronger and more smart, the parsing should be done better and better. We are currently using Claude Sonnet 4.5 which is the smartest model in the world. But of course it’s not perfect, so I’m always open for suggestions. I’m available @flaze9 on Discord if you want to talk.
That’s a super solid breakdown of the pros and cons! I’ve been feeling the exact same way about LR’s potential vs. its actual development speed.
The LR + asbplayer combo you mentioned is genius, though. I never thought about just using LR for the dictionary/UI and letting ASBplayer handle the one-click audio export to Anki. That basically solves the biggest pain point. I’m definitely going to try that setup this weekend.
You can explore the free tool as well. https://wordimageextractor.com/
