md-llm-lang-reader 0.2.0

Generate language-learning HTML readers (with sentence-level LLM translations and TTS) from Markdown.

langreader

Generate interactive language-learning HTML readers from Markdown using LLMs.

👀 Demo

Before reading further, check out what this tool generates:

👉 Live Demo: French Reader (Alsace)

---

md-llm-lang-reader (CLI command: langreader) converts standard Markdown text into a bilingual HTML reader. It uses Large Language Models (via multiai) to intelligently split paragraphs into natural sentences and provide translations, while adding Text-to-Speech (TTS) capabilities to every sentence.

Features

• 🤖 AI-Powered Translation: Automatically splits paragraphs into natural sentences and translates them line-by-line.
• 🔊 Text-to-Speech: Built-in audio buttons for every source sentence (uses browser Web Speech API).
• ⏯️ Resumable Generation: Hit an API rate limit? Use --continue to resume generation exactly where it stopped.
• 📝 Markdown Support: Preserves headings, code blocks, and basic formatting.
• 🌍 Wikipedia Helper: Includes a script to easily fetch and format Wikipedia articles for learning.

📚 Documentation

For detailed installation instructions, CLI options, and tutorials, please visit the official documentation:

👉 https://sekika.github.io/langreader/

• Usage Guide: Options for styling, resuming, and metadata.
• Tutorial: Step-by-step guide to creating a reader from a Wikipedia article.
• TTS Setup: How to configure high-quality voices on your device.

Installation

pip install md-llm-lang-reader

Note: This tool requires multiai configuration to access LLM providers (OpenAI, Anthropic, etc.).

Quick Start

1. Create a Markdown file (input.md):

   # Bonjour
   Ceci est un exemple de phrase pour l'apprentissage.
   

2. Run the generator:

   langreader -i input.md -o output.html --src fr --tgt en --provider openai --model gpt-4o-mini
   

3. Open output.html in your browser to start reading and listening!

License

MIT