ssnangua/ColorTxt

🎨 ColorTxt – “彩读” 3.0

What it is – A cross‑platform desktop reader (Electron + Vue + Monaco) for Chinese novels and other e‑books. It loads plain‑text, markdown or common e‑book formats, converts them to markdown, and then displays the content with a rich set of reading‑aid features.

Key capabilities

Feature	What you get
Local file support	Open .txt, .md and many e‑book formats (.epub, .mobi, .pdf, …). Files are converted to markdown for rendering.
Automatic chapter detection	Built‑in regex rules plus user‑defined patterns to build a clickable chapter list.
Content colouring	Custom highlight rules colour characters, keywords or any user‑chosen words, giving a “colored‑reading” experience.
Annotations	Bookmarks with notes, line‑level highlights, free‑form notes, and a “sticky” chapter title bar.
Editing & smart formatting	In‑place editing (fix typos) and an AI‑powered smart layout tool that can clean up hard line‑breaks, punctuation, HTML remnants, watermarks, etc., showing a diff before applying changes.
Reading helpers	Timed auto‑scroll, Pomodoro timer, full‑screen mode, and a quick‑hide “stealth” hot‑key.
Voice playback	Text‑to‑speech via system voices, Edge TTS, or cloud services (Aliyun Qwen3‑TTS, MiniMax, Xiaomi MiMo, etc.). Supports single‑voice narration or multi‑voice dialogue using AI‑generated character voice profiles.
AI Reading Assistant	Chat‑style interface that can:

• Analyse plot, answer story questions
• Generate mind‑maps and word‑clouds
• Create character cards (summaries + AI‑generated portrait images) using local or remote text‑to‑image models
• Perform RAG‑style full‑text retrieval with vector embeddings (local BGE‑Small or multilingual‑E5 models, or any OpenAI‑compatible embedding API) | | Translation & dictionary | Inline translation via many services (OpenAI, Microsoft, Google, DeepL, Baidu, etc.) and dictionary lookup (Wiktionary/Wikipedia plus user‑imported StarDict/MDict/DICT/Slob/BGL files). | | Book‑pack (彩读书包) | Export/import a .ctz (plain zip) or encrypted .ctzx bundle containing the original text, highlights, notes, bookmarks, character assets and config – handy for sharing or syncing across devices. | | WebDAV sync | Sync settings and book‑packs to a remote WebDAV server, enabling multi‑device usage. | | Cross‑platform | Runs on Windows, macOS and Linux (global hot‑key works on Windows/macOS; Linux Wayland has a known limitation). |

AI & model integration

• Chat models: any OpenAI‑compatible endpoint (OpenAI, Azure, Ollama, LM Studio, DeepSeek, DashScope, GLM, Moonshot, SiliconFlow, Agnes AI, MiniMax, Xiaomi MiMo, OpenRouter, Google Gemini, etc.).
• Embedding models: same endpoints plus built‑in Chinese BGE‑Small (≈47 MB) and multilingual E5‑Small (≈118 MB) for local RAG.
• Text‑to‑image: local WebUI/ComfyUI or cloud services (OpenAI Images, Stability AI, DashScope, MiniMax, Agnes AI) for generating character illustrations.

Why it matters – ColorTxt blends a conventional e‑book reader with modern AI utilities: contextual analysis, automatic formatting, multilingual translation, and voice narration with character‑specific voices. It lets readers not only consume a novel but also explore its structure, visualize characters, and interact with the story via AI‑driven tools, all while staying offline for most core functions.

Getting started – Install the pre‑built binaries (or run the source with npm/yarn), open a text or e‑book file, and explore the side panel for chapters, highlights, notes, and the AI assistant. Configure your preferred LLM/embedding endpoints in Settings → AI if you want remote capabilities; otherwise the bundled small models work out‑of‑the‑box.

---

All details above are taken directly from the repository’s README.