kenkui 0.5.0

Convert Ebooks to Audiobooks with [custom] voice samples

Kenkui

Kenkui is basically a fancy wrapper for Kyutai's pocket-tts, with support for ebook parsing. It is multithreaded, and runs faster than any other tool I've used, so I figured I'd start a project to make it easier to use.

Features

• Freaky fast audiobook generation
• No GPU needed, 100% cpu
• Super High Quality Text-to-Speech
• State of the Art Tools
• Multithreaded
• Custom voices
• Chapter selection
• Batch processing
• Automatic cover embedding from EPUB to M4B
• Fix existing audiobooks (salvage chapters, add missing ones, fix metadata)

Requirements

• python 3.7+
• pip, pipx, or uv
• ffmpeg
• mutagen (installed automatically)

Quick Start

Dependencies:

uv tool install kenkui
kenkui <your ebook name>.epub

You can also install using pip or pipx!

pip install kenkui

pipx install kenkui

Usage

You can pass a file or directory into kenkui, and it will search directories recursively for .epub files and convert them to m4b files.

Basic Examples

# Convert a single book
kenkui book.epub

# Convert entire library directory
kenkui library/

# Specify output directory
kenkui book.epub -o output/

# Use specific voice
kenkui book.epub --voice alba

Voice Selection

Use --voice to specify the voice you want to use.

• accepts:
  • the eight default voices of pocket-tts (alba, marius, javert, jean, fantine, cosette, eponine, azelma)
  • .wav files locally on computer (kenkui ships with quite a few extra)
  • voices from Hugging Face (hf://user/repo/voice.wav)

Use --list-voices to see all available voices.

Chapter Selection

Use --select-chapters to interactively choose which chapters to convert.

Fix Audiobooks

Use --fix-audiobook to fix existing M4B files:

# Scan audiobook for missing chapters and fix metadata
kenkui --fix-audiobook book.epub book.m4b

# With verbose output
kenkui --fix-audiobook book.epub book.m4b --verbose

What --fix-audiobook does:

1. Scans for missing chapters - Compares EPUB table of contents with M4B chapters
2. Salvages existing audio - Extracts matching chapters from existing M4B (no regeneration needed)
3. Generates missing chapters - Asks to generate any chapters not found in M4B
4. Rebuilds audiobook - Stitches all audio in proper EPUB chapter order
5. Embeds cover - Extracts cover from EPUB and embeds into fixed M4B
6. Creates new file - Outputs book_fixed.m4b (original preserved as backup)

Custom voices

In order to use your own custom voice, make sure to record a 5-10 second clip of the person speaking, with minimal background noise or crosstalk. We highly recommend using some sort of tool to clean the audio.

Examples

# Basic conversion
kenkui book.epub

# With voice selection
kenkui library/ --select-books --voice alba
kenkui book.epub --select-chapters --voice ~/Downloads/voice.wav

# With output directory and workers
kenkui book.epub -o output/ -w 4

# Fix an existing audiobook
kenkui --fix-audiobook book.epub book.m4b --verbose

Notes

At this time we do not plan on supporting mp3, not because it's hard, but because m4b is a wonderful format. There is also currently not support for ebook formats other than epub. I'm sure it'll get added in the future.

For similar reasons, currently only pocket-tts is supported as the tts provider. It's the smallest, fastest, and most feature complete at the moment.

Special Thanks

Thank you to the Guttenberg Project for providing some books included in kenkui!

Changelog

v0.4.0

• Automatic cover embedding - Covers from EPUB are now automatically embedded into M4B output
• Fix audiobooks - New --fix-audiobook command for:
  • Detecting missing chapters via fuzzy title matching
  • Salvaging existing audio from M4B files
  • Generating only missing chapters
  • Fixing metadata and embedding covers
• Fuzzy chapter matching - Smart matching between EPUB and M4B chapter titles
• Chapter salvage - Reuse existing audio instead of regenerating when possible