All projects

Cypher AI Book

Turns PDFs, Word files and EPUBs into natural-sounding audiobooks, locally.

WindowsmacOSLinuxSource on GitHub
View on GitHub Source and releases For this device

What Cypher AI Book is

A text-to-speech reader that runs on your own machine. Type or paste text, or upload a PDF, Word file or EPUB and let it pull the text out, then pick one of four speech models and hear it read aloud. VITS alone has 109 voices. A FastAPI backend does the synthesis on PyTorch, on the GPU when there is one, and a small web interface streams the audio back as it is made.

Cypher AI Book features

What it actually does.

Four speech models

Tacotron2 and GlowTTS for quality, SpeedySpeech for speed, and VITS for choice, voiced through a HiFiGAN vocoder.

109 voices

The multi-speaker VITS model offers 109 speakers. The other models read in one default voice.

Documents in, text out

Upload a PDF, DOCX or EPUB and its text is extracted automatically, ready to read.

Streaming playback

Audio plays through the Web Audio API as it arrives, instead of after the whole file is finished.

GPU when you have one

Runs on CUDA with an NVIDIA card and falls back to the CPU without one. Nothing is sent to a cloud service.

A plain HTTP API

Models, voices, synthesis, document upload, system information and a health check are all endpoints you can call yourself.

How to use Cypher AI Book

From a fresh install to the thing working.

1 Install it

  1. You need Python 3.10+, Node.js 14+ and at least 4 GB of RAM (8 GB is better). A GPU is optional.
  2. Create an environment (conda is recommended, because it handles CUDA better) and pip install -r requirements.txt. For a GPU, install the CUDA build of PyTorch first.
  3. Run python setup.py to download the models ahead of time, or let each one download the first time you use it. Together they come to 1 to 3 GB.
  4. Run npm install for the interface.

2 Start it

  1. In one terminal, python run.py starts the backend on port 9001.
  2. In another, npm start serves the interface on port 3100.
  3. Open http://localhost:3100.

3 Read something aloud

  1. Type or paste text, or upload a PDF, DOCX or EPUB.
  2. Choose a model, and a voice if that model has more than one.
  3. Press Synthesize. Playback starts as the audio streams in.

4 If there is no sound

  1. Check the backend answers: curl http://localhost:9001/health.
  2. Open the browser console (F12) and look for errors.
  3. If a model will not load, download it again by hand with the command in the README.

Built with

Python 3.10+ FastAPI PyTorch (CUDA or CPU) Coqui TTS models HiFiGAN vocoder PyPDF2, python-docx, ebooklib Node.js / Express Web Audio API

Licence

Made by Kencypher. Not open source.

Cypher AI Book was written by Kencypher. The source is not published and is not licensed for reuse. You have permission to run it — that is the whole of it.

You may

  • Run it on as many of your own machines as you like
  • Use it for anything, including commercial work
  • Pass an unmodified copy to a friend, free

You may not, without asking

  • Sell it or charge for access
  • Re-upload or mirror it anywhere
  • Rebrand it or publish it as your own
  • Decompile it or lift its assets

This is enforced. Re-uploading it, repackaging it, stripping the author's name off it or passing it off as your own is copyright infringement — protected without registration under the Berne Convention — and is pursued with DMCA takedowns to the host, CDN, app store and search engine, platform reports for impersonation, and legal proceedings where the infringement is commercial, repeated, or involves falsely claiming authorship.

Asking first is free, and the answer is usually yes — kencypher56@gmail.com. The only official source is this store; a copy from anywhere else is not one Kencypher published.

Good to know