k8bit/alexandria-audiobook repository overview

https://github.com/Finrandojin/alexandria-audiobook⁠

Alexandria Audiobook Generator

Transform any book or novel into a fully-voiced audiobook using AI-powered script annotation and text-to-speech. Features a built-in Qwen3-TTS engine with batch processing and a browser-based editor for fine-tuning every line before final export.

Features AI-Powered Pipeline Local & Cloud LLM Support - Use any OpenAI-compatible API (LM Studio, Ollama, OpenAI, etc.) Automatic Script Annotation - LLM parses text into JSON with speakers, dialogue, and TTS instruct directions LLM Script Review - Optional second LLM pass that fixes common annotation errors: strips attribution tags from dialogue, splits misattributed narration/dialogue, merges over-split narrator entries, and validates instruct fields Smart Chunking - Groups consecutive lines by speaker (up to 500 chars) for natural flow Context Preservation - Passes character roster and last 3 script entries between chunks for name and style continuity Voice Generation Built-in TTS Engine - Qwen3-TTS runs locally with no external server required External Server Mode - Optionally connect to a remote Qwen3-TTS Gradio server Multi-Language Support - English, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Russian, Spanish, or Auto-detect Custom Voices - 9 pre-trained voices with instruct-based emotion/tone control Voice Cloning - Clone any voice from a 5-15 second reference audio sample Voice Designer - Create new voices from text descriptions (e.g. "A warm, deep male voice with a calm and steady tone") LoRA Voice Training - Fine-tune the Base model on custom voice datasets to create persistent voice identities with instruct-following Built-in LoRA Presets - Pre-trained voice adapters included out of the box, ready to assign to characters Dataset Builder - Interactive tool for creating LoRA training datasets with per-sample text, emotion, and audio preview Batch Processing - Generate dozens of chunks simultaneously with 3-6x real-time throughput Codec Compilation - Optional torch.compile optimization for 3-4x faster batch decoding Non-verbal Sounds - LLM writes natural vocalizations ("Ahh!", "Mmm...", "Haha!") with context-aware instruct directions Natural Pauses - Configurable silence between speakers (default 500ms) and same-speaker segments (default 250ms) Web UI Editor Streamlined Interface - 5-step core pipeline (Setup, Script, Voices, Editor, Result) plus advanced tools (Designer, Dataset, Training) Chunk Editor - Edit speaker, text, and instruct for any line Selective Regeneration - Re-render individual chunks without regenerating everything Batch Processing - Optimized batch rendering with sub-batching for efficient GPU utilization Live Progress - Real-time logs and status tracking for all operations Audio Preview - Play individual chunks or sequence through the entire audiobook Script Library - Save and load annotated scripts with voice configurations Export Options Combined Audiobook - Single MP3 with all voices and natural pauses Individual Voicelines - Separate MP3 per line for DAW editing (Audacity, etc.) Audacity Export - One-click zip with per-speaker WAV tracks, LOF project file, and labels for automatic multi-track import into Audacity M4B Audiobook - Chaptered M4B (AAC) with per-chunk or auto-detected chapter markers for audiobook players (Audiobookshelf, Apple Books, VLC, etc.) Requirements Pinokio LLM server (one of the following): LM Studio (local) - recommended: Qwen3 or similar Ollama (local) OpenAI API (cloud) Any OpenAI-compatible API GPU: 8 GB VRAM minimum, 16 GB+ recommended — see compatibility table below Each TTS model uses ~3.4 GB; remaining VRAM determines batch size CPU mode available on all platforms but significantly slower RAM: 16 GB recommended (8 GB minimum) Disk: ~20 GB (8 GB venv/PyTorch, ~7 GB for model weights, working space for audio)