Alexandria transforms books into audiobooks by leveraging LLMs to parse text into annotated scripts with speaker attribution, dialogue tags, and TTS style instructions. It then uses Qwen3-TTS for text-to-speech generation with support for voice cloning and LoRA model training to create custom voices. The system supports per-line style control and exports to MP3, chaptered M4B, or multi-track Audacity formats, with a browser-based editor for fine-tuning before final rendering.