PDF2MP3 – Convert PDFs into Audiobooks

A command-line utility (CLI) in Python to convert PDF files into MP3 audiobooks using high-quality, natural female neural voices.
Supports English (en) and Brazilian Portuguese (pt-br).

---

Features

• Extracts text from PDFs using pdfminer.six.
• Converts text to speech using edge-tts (Microsoft Neural Voices).
• Default high-quality female voices:
  • en-US-AriaNeural
  • pt-BR-ThalitaNeural
• Smart chunk splitting for fluent narration (default: 1600 characters per chunk).
• Generates a single continuous MP3 file (via pydub + ffmpeg).
• Configurable language, voice, speaking rate, volume, and chunk size.
• Built-in retry system with exponential backoff for long or unstable requests.

---

Installation

1. Clone the repository

git clone https://github.com/byraphaelmedeiros/pdf2mp3.git
cd pdf2mp3

2. Create a virtual environment (recommended)

python -m venv .venv
.venv\Scripts\activate   # Windows
# source .venv/bin/activate  # Linux/Mac

3. Install dependencies

pip install --upgrade pip
pip install -r requirements.txt

4. Install FFmpeg

pydub requires ffmpeg to export audio to MP3.

Windows (via winget):

winget install --id=Gyan.FFmpeg -e

macOS (via Homebrew):

brew install ffmpeg

Linux (Debian/Ubuntu):

sudo apt-get update && sudo apt-get install ffmpeg

Verify the installation:

ffmpeg -version

---

Usage

Basic usage requires only the input PDF. Other parameters have defaults.

python pdf2mp3.py <input.pdf> [options]

Parameters

• Required
  • input → Input PDF file
• Optional
  • -o, --output → Output MP3 file (default: same name as PDF with .mp3)
  • -l, --lang → Language (en or pt-br, default: pt-br)
  • --voice → Voice name (default: AriaNeural / ThalitaNeural)
  • --rate → Speaking rate (e.g., +5%, -5%, default: +0%)
  • --volume → Speaking volume (e.g., +0%, +5%, default: +0%)
  • --max-chars → Maximum characters per chunk (default: 1600)

---

Examples

Convert a PDF in Portuguese (minimal usage):

python pdf2mp3.py "document.pdf"

Output: document.mp3 in pt-br using pt-BR-ThalitaNeural.

Convert a PDF in English:

python pdf2mp3.py "book.pdf" -l en

Specify a custom output file:

python pdf2mp3.py "document.pdf" -o "output.mp3"

Adjust speaking rate and volume:

python pdf2mp3.py "document.pdf" --rate +5% --volume +0%

Use smaller chunks:

python pdf2mp3.py "document.pdf" --max-chars 1200

---

Notes

• This program does not work with scanned PDFs (image-based).
  Use OCR first, for example:

  pip install ocrmypdf
  ocrmypdf --force-ocr input.pdf output_ocr.pdf
  

• Audio quality depends on the text extraction results from pdfminer.six.

• An active internet connection is required (the edge-tts library uses Microsoft’s neural voices).

---

Quick Checklist

1. Install Python dependencies:

   pip install -r requirements.txt
   

2. Install FFmpeg:

   winget install --id=Gyan.FFmpeg -e   # Windows
   

3. Run the program:

   python pdf2mp3.py input.pdf
   

You will get a ready-to-play MP3 audiobook.

---

Contributing

Contributions are welcome!
Please read the Contributing Guidelines for details.

---

Code of Conduct

This project follows a Code of Conduct.
By participating, you are expected to uphold this standard.

---

Security

If you discover a security vulnerability, please report it by emailing:
pdf2mp3@byraphaelmedeiros.com
See SECURITY.md for more details.

---

License

This project is released under the terms of the MIT License.
See the LICENSE file for details.

---

Maintainer

Raphael Medeiros

• GitHub: @byraphaelmedeiros
• Contact: pdf2mp3@byraphaelmedeiros.com