Keep the author's words
Retain EPUB reading order, source locators, headings, and paragraph lineage. Record justified changes to the speech surface.
LOCAL TEXT-TO-SPEECH · CODEX PLUGIN
EPUB to audiobook.
Make the words act.
AI Audiobook Generator turns EPUB and HTML into expressive English audiobooks with Chatterbox TTS, reference voices, and chaptered M4B export. Preserve the edition, direct its performance, and review the finished book.
Development preview · Python 3.13 · English · Local execution
AI audiobook generation with source fidelity
Direct the listener's experience
Score the argument before synthesis: clarify, probe, convince, warn, invite, exalt, urge, and resolve. Keep direction out of spoken text.
Export chaptered M4B
Resume verified audio segments, assemble lossless samples in order, and derive chapter positions from realized audio.
DIRECTED ORAL INTERPRETATION
Give the argument a living voice.
Some writing addresses a listener: it questions, invites, warns, persuades, or exhorts. Expressive narration can make those actions audible while preserving the text and its qualifications.
The director makes an edition-specific rhetorical score. Adaptation prepares faithful speech, Chatterbox renders the accepted performance, and editorial and listening review determine whether the delivery is ready.
Read the business narrative, sources, and acting framework ↗
GET STARTED
Install the AI audiobook generator in Codex
codex plugin marketplace add https://github.com/fredfortier/ai-audiobook-generator.git
codex plugin add ai-audiobook-generator@ai-audiobook-generatorStart a new Codex session. The plugin contains three skills: preprocess-text, adapt-text, and chatterbox.
Provision Python 3.13, uv, FFmpeg/ffprobe, English NLTK data, and an original Chatterbox model snapshot. A compatible CUDA GPU is recommended for production. Keep books and exports outside the plugin cache.
Follow the complete setup instructions ↗
Use preprocess-text to prepare my EPUB, adapt-text for faithful
speech and reviewed rhetorical direction, then chatterbox for
auditions and a reviewed chaptered M4B audiobook.Before your first audiobook
Which books can it process?
The supported intake is EPUB or HTML, with English narration using original Chatterbox. The agent resolves edition scope and ambiguous source structure before adaptation.
Does it clone a reference voice?
Chatterbox uses reference-audio conditioning. The bundled Enoch recording is CC BY 4.0; source books and other recordings retain their own rights and terms.
Is this a one-click converter?
The workflow includes editorial decisions, auditions, and listening review. Automated validation establishes structure and content identities; it cannot certify acting or pronunciation.
What is the current release status?
This is a development preview. Strict type checking has a known backlog. See current checks and tracked issues before using it for production.