PDF to Audio Converter using Azure, AWS, Eleven Labs, or OpenAI Text-to-Speech

This project converts PDF files into high-quality audio files using Azure, AWS, Eleven Labs, or OpenAI Text-to-Speech services. It extracts text from PDFs, processes it, and generates audio output, making it ideal for creating audiobooks or listening to documents on the go.

Table of Contents

• Features
• Prerequisites
• Installation
• Setup
• Usage
• Configuration
• Notes
• Troubleshooting
• Dependencies
• Contributing
• License

Features

• Support for multiple Text-to-Speech providers: Azure, AWS, elevenlabs, and OpenAI
• Batch processing of multiple PDF files
• Automatic text chunking for efficient processing
• Customizable voice and language settings
• Error handling and logging

Prerequisites

• An active account with one of the following services:
  • Azure Cognitive Services
  • AWS Polly
  • Eleven Labs
  • OpenAI
• API credentials for the chosen service
• Python 3.12 or higher
• pip 23.0 or higher

Installation

1. Clone the repository:

   git clone https://github.com/JaffeMarques/PDFtoSpeechAI.git
   cd PDFtoSpeechAI

2. Set up a virtual environment:

   python3 -m venv myenv
   source myenv/bin/activate  # On Windows, use `myenv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

Setup

1. Create required directories:

   mkdir input chunks output

2. Configure environment variables: The project includes an env.example file with all the necessary environment variables. To set up your configuration:

   a. Copy the env.example file to .env:

   cp env.example .env

   b. Open the .env file and fill in your credentials and preferences:

   # Choose your provider (azure, aws, elevenlabs, or openai)
   TTS_PROVIDER=
   
   # General settings
   MAX_CHUNK_SIZE=1000
   SLOW_DOWN_RATE=
   
   # Eleven Labs credentials
   ELEVENLABS_SPEACH_SECRET=
   ELEVENLABS_SPEACH_URL=
   ELEVENLABS_SPEACH_VOICE_ID=
   
   # OpenAI credentials
   OPENAI_SPEACH_SECRET=
   OPENAI_SPEACH_VOICE=
   OPENAI_SPEACH_MODEL=tts-1
   
   # Azure credentials
   AZURE_SPEACH_SECRET=
   AZURE_SPEACH_REGION=
   AZURE_SPEACH_VOICE=
   AZURE_SPEACH_LANGUAGE=
   
   # AWS credentials
   AWS_SPEACH_KEY=
   AWS_SPEACH_SECRET=
   AWS_SPEACH_REGION=
   AWS_SPEACH_VOICE=
   AWS_SPEACH_LANGUAGE=
   

   Fill in the appropriate values for your chosen provider and preferences. You only need to fill in the credentials for the provider you're using.

Usage

1. Place PDF files in the input folder.

2. Run the conversion script:

   python main.py

3. Find the generated audio files in the output folder.

Configuration

You can customize the following settings in the .env file:

• TTS_PROVIDER: Choose between 'azure', 'aws', 'elevenlabs', or 'openai'
• MAX_CHUNK_SIZE: Maximum number of characters per text chunk
• SLOW_DOWN_RATE: Adjust the speech rate (if supported by the provider)
• Provider-specific settings like voice, language, and region

Notes

• This project is designed for educational purposes and may not be suitable for production environments without further optimization.
• Be mindful of service limits and quotas to avoid throttling or additional charges.
• Large PDFs may take considerable time to process.
• The script includes basic error handling, but you may need to implement additional checks for specific use cases.

Troubleshooting

• If you encounter "File not found" errors, ensure that the input, chunks, and output folders exist in the project root.
• For API-related issues, verify your credentials in the .env file and check your account status with the chosen provider.
• If the audio quality is not satisfactory, try adjusting the MAX_CHUNK_SIZE or experimenting with different voice settings.

Dependencies

See requirements.txt for a full list of dependencies.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.