chromewebstore.google.com
Show HN: CastReader – Free TTS Extension That Reads Kindle Cloud Readerhttps://chromewebstore.google.com/detail/castreader-tts-reader/foammmkhpbeladledijkdljlechlclpbEvery TTS browser extension fails on Kindle Cloud Reader. The reason: Amazon renders text using custom font subsets where glyph IDs don't map to standard Unicode. You select text, copy it, and get garbage. The DOM is useless. CastReader solves this by intercepting KindleModuleManager to capture font and token data, decoding glyph mappings from the binary font tables, then running Tesseract.js OCR locally in an offscreen document to calibrate the decoder. The final text comes from glyph decoding (not OCR) so it's accurate enough for word-level highlight sync. WeRead (the largest Chinese reading platform) has a similar problem — it renders everything on canvas. CastReader uses a main-world content script injected at document_start to intercept fetch responses containing chapter data before the page consumes them. For normal websites, there's a 3-tier extraction pipeline: 15+ site-specific extractors (Notion, Google Docs, ChatGPT, Claude, arXiv, etc.), a learned CSS selector rule system, Mar 12, 2026 7:07 PM