DramaLex Learning English From TV Series And Films

Problem

Watching series or films often fails as a study method because input, production, and review stay disconnected. Learners may understand every subtitle, still miss natural speech; remember isolated words, but fail to use them in speaking or writing; finish one episode, then lose the material because review is fragmented. DramaLex turns one episode into a closed loop: it calibrates level first, then recycles the same target vocabulary, patterns, and lines across listening, reading, speaking, and writing.

How It Works

• Input: provide a title and episode, such as Friends S01E01, or a film title; the agent searches public subtitle sources, parses .srt/.vtt files, and checks the version.
• Diagnosis: diagnose.json maps IELTS, TOEFL, CET scores or an adaptive quiz to a CEFR level, which calibrates task difficulty.
• Generation: the workflow moves from Phase 0 to Phase 6, producing preview cards, a viewing protocol, listening comprehension and dictation, transcript annotations, shadowing/role-play tasks, and rewrite/continue/summary writing tasks.
• Feedback: speaking can use score_speaking.py for Whisper transcription comparison; writing can use score_writing.py for has_word, min_words, max_words, and tense checks before agent correction.
• Review: vocab_bank.json tracks previously learned items, and HTML or Anki supports spaced repetition.

Boundaries

The skill is aimed at English learners with a Chinese-language learning context. It depends on network access and public subtitle sources, and it is intended for personal non-commercial study; it does not host or distribute subtitles. Speaking scoring compares transcription to the target line and does not rate accent. Full results still depend on watching the episode, completing output tasks, and reviewing consistently.