Converts non-DRM e-books and documents into audiobooks with chapters and metadata using text-to-speech. Supports multiple TTS engines, custom models, voice cloning, and synthesis across 1158 languages and dialects.
Project overview
Supports text-to-speech synthesis in 1158 languages and dialects and is documented to run on hardware as limited as 2 GB RAM and 1 GB VRAM.
Project type
Document Processing · Audio & Speech
Use cases
Meeting Notes · Audio & Speech
Deployment
Refer to project documentation
License
Apache-2.0
Best for
General users and creators who need to convert text or legally acquired, non-DRM documents into audio formats.
Key capabilities
Converts multiple e-book and document file formats (.epub, .mobi, .pdf, .txt, etc.) into audio formats (.m4b, .mp3, .wav, etc.) with chapters and metadata.
Provides optional voice cloning using a user-provided audio file, including built-in de-noising for the reference audio.
Supports text-to-speech synthesis in 1158 languages and dialects.
Supports synthesis using XTTSv2, Bark, Fairseq, VITS, Tacotron2, Tortoise, GlowTTS, and YourTTS.
Performs OCR scanning for e-book files that contain text pages as images.
Allows direct conversion of short text to audio via a TextArea.
Optionally translates e-books to a target language using argostranslate before TTS synthesis.
Provides fine-grained control over breaks, pauses, and voice switching via SML tags.
Limitations and risks
TTS engines are very slow on CPU compared to GPU.
EPUB format lacks standard structure, which may require manual removal of unwanted text before conversion.
MPS (Apple Silicon) is not exposed in Docker, requiring CPU usage in that environment.
The tool is intended for non-DRM, legally acquired eBooks only.
Getting started
Setup is documented as easy, providing double-click binaries for Mac and Windows, or repository scripts. The documented first success path is to clone the repository, run ./ebook2audiobook.command, and open http://localhost:7860/.
Alternatives and comparisons
Converts PDFs, EPUBs, and subtitles into audiobooks using local models and voice cloning.
Provides a WebUI for zero-shot and few-shot voice conversion and TTS with minimal voice data requirements.
Offers local-first control over clinical audio transcription and note generation.
README: CPU/GPU Converter from E-Book to audiobook with chapters and metadata<br/> using advanced TTS engines and much more.<br/> Supports voice cloning and 1158 languages!