QuickPiperAudiobook

With one command, create a natural-sounding audiobook from a variety of input formats (epub, mobi, txt, PDF, HTML and more!)

Last updated Jul 8, 2026

1.1k

Stars

Forks

Issues

Stars/day

Attention Score

View on GitHub ↗ Open in github.dev

Topics

#audiobooks #piper #productivity #tts

Language breakdown

No language data available.

▸ Files click to expand

README

QuickPiperAudiobook

English | Polski | 简体中文

Create a natural audiobook for any text content with one command.

- Converts PDFs, epub, txt, mobi, djvu, HTML, docx, and more - All conversion is done offline and is entirely private - Uses piper models; supports many languages

Listen to sample output here

Installing

Download the binary via one of the following methods:

* A prebuilt release * go install github.com/C-Loftus/QuickPiperAudiobook@latest * (Or build from source using go build)

Download ebook-convert and make sure it is in your PATH. (This is bundled with calibre)
(Optional) Download ffmpeg for mp3 and chapter support

[!NOTE]

You don't need to have piper installed. This program manages piper and the associated models

Usage

Pass in either a local file or a remote URL with the proper extension

* i.e. ./QuickPiperAudiobook test.txt

Specify the --chapters flag to generate mp3 chapters for epub files

* i.e. ./QuickPiperAudiobook --chapters test.epub

For a full list of options use the --help flag

* i.e. ./QuickPiperAudiobook --help

Non-English / UTF-8

Grab a model for your language of choice (.onnx and .json) from the piper models

* i.e. plPL-gosia-medium.onnx and corresponding plPL-gosia-medium.onnx.json (rename if needed)

Put them in ~/.config/QuickPiperAudiobook/
Use the --speak-utf-8 and --model= flags to specify you want utf characters to be spoken with a specific model

* i.e. ./QuickPiperAudiobook --speak-utf-8 --model=plPL-gosia-medium.onnx MaszynaTuringaWikipedia.pdf

[!NOTE]

Consider specifying this model as the default in the configuration file if you plan to use it frequently

Configuring

You can create a config file at ~/.config/QuickPiperAudiobook/ to specify preferred values if you do not want to specify these as cli args each time

* i.e. you can use any arbitrary model by putting the associated .onnx and .onnx.json file for it in ~/.config/QuickPiperAudiobook/ * A full example config can be found here

# An example for ~/.config/QuickPiperAudiobook/config.yaml
the default output directory to use if the user does not specify --output in the cli args
output: ~/Audiobooks
the default model to use if the user does not specify --model in the cli args
model: "enUS-hfcfemale-medium.onnx"
output the audiobook as an mp3 file (requires ffmpeg in your PATH)
mp3: false
generate chapter metadata when outputting mp3s (requires an epub input and ffmpeg in your PATH)
chapters: false

Notes

Piper does not support progress output. Long audiobooks may take a long time to generate since all computation is being done locally.
This repo has been tested on Linux but can also work on MacOS once an upstream bug is fixed with Piper

Support

Thank you for considering supporting this project.

I accept donations on Github or Paypal. If you would like to sponsor this project or reach out to me for business reasons, you can contact me via email or my website

🔗 More in this category