StudioDeveloper releases fine-read-it-myself, a local multi-voice audiobook producer
Original title: Fine, I’ll read it myself - local audiobook producer
The author released fine-read-it-myself, a desktop app that turns web novels into multi-voice audiobooks with a different voice per character, running entirely locally with no subscription. It ships with plugins for Royal Road, Wuxiaworld and NovelFire, with declarative YAML plugins for more sites. Chapters go through three inspectable steps: ModernBookNLP/BookNLP (a C++/ggml port) splits narration and quotes and detects speakers; a local LLM (Qwen3.5-4B, Qwen3.5-9B or Gemma 4 12B via llama.cpp, or any OpenAI-compatible endpoint) double-checks speaker labels and flags uncertain cases; VoxCPM2 (via audio.cpp) reads the script into tagged MP3s using persistent voices from a pool of LibriVox readers. Everything runs on Vulkan (NVIDIA, AMD, Intel), the UI is shareable over the local network, …
The author open-sourced a local multi-voice audiobook producer with inspectable, editable steps for text analysis, speaker verification and narration, useful for web-novel listeners.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.