Show HN: Curious about tones in Chinese? An extension for language learners (github.com)

A new open‑source Chrome extension makes reading Chinese on any webpage more learner‑friendly by color‑coding characters by tone, offering a pop‑up CEDICT‑based dictionary, side‑panel definitions, example sentences (from Tatoeba), and quick links to pronunciation resources like Forvo and Youglish. It tokenizes text using the browser Intl APIs, displays tone colors on demand, and can run sentence/word analyses via optional OpenAI API integration (gpt-4o for text, tts-1 “nova” for speech). The extension also hooks into Anki through AnkiConnect for one‑click flashcard creation (cards are auto‑tagged tag:ChineseLearningExtension), supports audio cards, and uses structured LLM outputs to keep AI responses predictable. For the AI/ML and language‑tech communities this is notable because it blends lightweight browser tokenization, offline resources (CEDICT/Tatoeba), and LLM augmentation in a privacy‑aware, bring‑your‑own‑key model. Practical implications include rapid creation of spaced‑repetition material from real web content, AI‑assisted grammatical breakdowns for learners, and a pragmatic solution to missing third‑party audio (Forvo’s CORS removal pushed the author toward TTS). Future work calls out richer media types, improved prompt engineering, and possibly local proxying for blocked APIs—making this a useful prototype for research and tooling that combines client‑side NLP, LLM enhancements, and study workflow automation.