Unihongo

Language apps routinely collapse at the speaking step — flashcards and reading drills are easy to ship; realistic conversation practice is not. Unihongo centers its product on an AI speaking practice module where learners work through 48+ topic scenarios drawn from daily life, travel, and work contexts. You pick a teacher personality, enter a 3D classroom environment, and practice under real-time correction. The skill-building section organizes exercises by JLPT level (N5 through N1) and category, including grammar, vocabulary, listening, kanji, and reading, with community-contributed questions and explanations.

The Speed Test feature is the differentiating mechanic for retention under pressure. Timed quiz rooms serve random questions from the question bank, community rankings track comparative progress, and badge milestones give short-loop motivation signals. For learners who plateau on passive study, the time pressure and public ranking add a layer of accountability that solo flashcard review does not.

The flashcard system is tied directly to in-app content — when you encounter an unfamiliar word in a question, you select it, look up the meaning, and add it to your deck. Organizing cards by day or topic makes spaced review manageable. This integration keeps vocabulary work connected to the grammar and reading contexts where you found the words, rather than treating vocab as a separate silo.

Unihongo is a closed, hosted consumer application. There is no API, no self-hosted deployment option, and no documented export or integration layer. For a self-studying learner building a daily JLPT prep routine, that is irrelevant. For a language school, a developer building a learning product, or anyone who needs progress data outside the app, the closed architecture is a hard stop.