ebook2audiobook reads an e-book and writes an audiobook. It takes EPUB, MOBI, AZW3, PDF, DOCX, TXT and a long list of other formats, runs OCR on scanned pages, splits the text into chapters, and narrates it with a local text-to-speech engine: XTTSv2 by default, or Bark, VITS, Fairseq, Tacotron2, Tortoise, GlowTTS or YourTTS. The result is an M4B (or MP3, FLAC, M4A and others) with chapter markers and metadata. Fairseq's models extend coverage to more than 1,100 languages, inline tags add pauses and voice switches, and you can clone a narrator from your own voice sample.
Drew Thomasson started it in January 2024 and it has about 20,000 stars, with a Gradio web interface, a headless command line and Docker images for CPU and several CUDA versions. It runs on as little as 2 GB of RAM and 1 GB of VRAM, though the better engines want a real GPU. The code is Apache-2.0, and the README states plainly that it is for DRM-free, legally acquired books only.
- Repository: github.com/DrewThomasson/ebook2audiobook
- Licence: Apache-2.0 (Apache License 2.0)
- Language: Python. Stars: 20.3K. Forks: 1,730. Last push: Oct 2, 2026.
- Scan: safe, Oct 2, 2026, commit d3ff5bc
Who it is for
Readers who want audio versions of books that have none, language learners, people with visual impairments or dyslexia, and anyone with a backlog of public-domain or self-published books.
Getting started
1. Clone the repository
git clone https://github.com/DrewThomasson/ebook2audiobook.git && cd ebook2audiobook2. Install and launch the web interface on Linux or macOS (Windows: ebook2audiobook.cmd), then open http://localhost:7860
./ebook2audiobook.command3. Or convert headless
./ebook2audiobook.command --headless --ebook mybook.epub --language eng4. Or run the CPU Docker image
docker run -v "./ebooks:/app/ebooks" -v "./audiobooks:/app/audiobooks" --rm -it -p 7860:7860 athomasson2/ebook2audiobook:cpuThe launcher installs missing tools for you: Homebrew on macOS and Scoop on Windows. Models download on first run. On a CPU, XTTSv2 and Bark are very slow; use a GPU or a lighter engine such as YourTTS or Tacotron2. The web interface listens on all network interfaces, so other devices on your network can reach it while it runs, and the --share flag goes further with a public Gradio link.
Safety scan
We cloned DrewThomasson/ebook2audiobook at commit d3ff5bc on Oct 2, 2026 and ran the checks described on the GitHub Tools page: credential patterns, decode-and-execute code, install-time scripts, committed binaries, risky CI workflows, every host the code talks to, known vulnerabilities in pinned dependencies, and project hygiene. A person read every hit. This is what we found.
- No secrets, no committed binaries and no bare-IP URLs across 339 files and about 49,000 lines. The three pattern hits are a Dockerfile installing Rust with rustup's curl | sh during an image build, a Windows launcher message telling you how to install Docker, and a long line of language-code mappings in lib/conf_models.py.
- lib/conf.py sets interface_host to 0.0.0.0, and app.py passes it to Gradio, so the converter is reachable from other devices on your network without a password. The --share flag goes further and publishes a temporary public gradio.live link.
- components/rocmfix.py is a vendored copy of ROCmFix, an AMD GPU helper. ebook2audiobook only calls its detection code, which downloads a GPU compatibility list from rocmfix-data.onrender.com when it meets an AMD card it does not know; ROCmFix's own opt-in telemetry is never called. The launch scripts install Homebrew on macOS or Scoop on Windows if they are missing.
- requirements.txt has 52 lines but pins only three packages (Gradio 6.28.0, coqui-tts and piper-tts), none with a known advisory; the other dependencies resolve to whatever is current at install time and could not be checked.
- Eight workflows, none using pull_request_target; neither of the 2 third-party actions is pinned. Licence and code of conduct present; no security policy, Dependabot, CodeQL or contributing guide.
What the scanner counted
| Check | Result |
|---|---|
| Secrets | None found. |
| Suspicious code | 3 pattern hits found and read; every one is listed under the raw findings. |
| Install-time code | None: nothing runs at install beyond the package manager itself. |
| Committed binaries | None. |
| CI workflows | 8 workflows. None use pull_request_target. 2 of 2 third-party actions pinned to a tag rather than a commit. |
| Network hosts | 18 distinct hosts referenced from source; most often github.com, colab.research.google.com, raw.githubusercontent.com, download.pytorch.org. No URLs to bare IP addresses. |
| Known vulnerabilities | 0 advisories across 3 pinned packages: 0 critical, 0 high, 0 moderate, 0 low. requirements.txt: 3 packages, 0 advisories. |
| Project hygiene | Has licence file. Missing security policy, automated dependency updates, CodeQL, contributing guide. |
| OpenSSF Scorecard | Not scored: the project is not in Scorecard's weekly index. |
The raw findings
Every hit the scanner wrote out, with a link to the exact line at the scanned commit. Secrets candidates are redacted.
Pattern hits (3)
| Where | Rule | Match |
|---|---|---|
| Dockerfile:148 | download-piped-to-shell | curl https://sh.rustup.rs -sSf | sh -s -- -y --default-toolchain stable; \ |
| ebook2audiobook.cmd:578 | download-piped-to-shell | echo Then manually run: curl -fsSL https://get.docker.com ^| sh |
| lib/conf_models.py:258 | very-long-line | 18109 chars |
By the numbers
| Stars | 20.3K |
|---|---|
| Forks | 1,730 |
| Contributors | 43 |
| Commits | 15.2K |
| Open issues | 2 |
| Open pull requests | 0 |
| Releases | 42 |
| Latest release | v26.9.27 |
| Licence | Apache-2.0 |
| Main language | Python |
| Project age | 2 years |
| Last push | Oct 2, 2026 |
| Tracked files | 339 |
| Lines of code | 49.2K |
| Checkout size | 36 MB |
Lines by language: Python 20.9K, Markdown 17.5K, JSON 5,983, YAML 1,571, Batch 1,248, CSS 685.
Questions
Is ebook2audiobook free?
The software is Apache-2.0 and free, with no API costs since everything runs locally. The voice models have their own terms: the default XTTSv2 model is under the Coqui Public Model License, which allows non-commercial use only, so selling audiobooks made with it needs a different engine or licence.
Can ebook2audiobook convert Kindle books?
It reads AZW3, MOBI and other Kindle formats, but only without DRM. Books bought from Kindle and most other stores are DRM-protected, and the project supports only DRM-free, legally acquired files. Public-domain sites such as Project Gutenberg are a good source.
How long does a book take to convert?
It depends on the engine, the book and your hardware. With XTTSv2 a full novel can take hours even on a good NVIDIA GPU and far longer on a CPU, which is why the README suggests lighter engines such as YourTTS or Tacotron2 for CPU use; they are faster but sound more robotic.
This post is part of GitHub Tools, where every repository is cloned and scanned before it is written up. The scan is a snapshot of one commit on one day; the repository has moved on since, so check it before you install.