Overview
Lexivine explores language practice through a mobile experience built around a voice-first conversation canvas with transcript cards. A language coach for adaptive voice roleplay, transcript feedback and practice built from the phrases you save. The proposed journey connects roleplay room, speech feedback, phrase deck, learning path. Four original phone showcases establish the visual direction, while the case study defines the supporting feature set and a proposed Flutter architecture.



Original mobile interface concepts with fictional content. AI responses illustrate proposed interactions; they are not live model results.
Challenge
Language learners need speaking practice that connects feedback to their own words rather than presenting a generic score. On a phone, the person also needs to capture information quickly, understand where an answer came from and return to the original material without navigating through a dense desktop interface.
- Keep roleplay room and speech feedback connected within a short mobile flow.
- Make adaptive roleplay correctable when the model misreads the input.
- Explain practice retrieval through visible references or user-selected preferences.
- Keep changes, sharing and data retention under explicit user control.
Solution
The design begins with a voice-first conversation canvas with transcript cards, using large touch targets, short task sequences and contextual sheets. Adjust fictional conversation turns to the selected level and topic. Suggest alternative wording beside the learner's original phrase. The companion views keep drafts, original material and accepted decisions distinct. A violet, white and lime visual system gives this app its own character within the collection.
- Adjust fictional conversation turns to the selected level and topic.
- Suggest alternative wording beside the learner's original phrase.
- Build review cards from phrases the learner explicitly saves.
- Offer draft pacing and pronunciation observations with audio replay.
Results
4
Original mobile
showcase views12
Designed
feature areas4
Proposed AI
workflows


