Here is the README for the audioSummarizer repository based on the provided code structure:

---

Audio Summarizer

The Audio Summarizer package converts PDF text into a podcast-ready audio format. This package extracts text from PDF files, processes the content using language models, and generates high-quality, engaging audio using Text-to-Speech (TTS) models.

Features

• PDF Text Extraction: Extracts text from PDF files using the pdfplumber library.
• Text Cleaning and Processing: Cleans and processes extracted text to make it more suitable for podcast narration.
• Podcast Script Generation: Uses Hugging Face and Ollama models to rewrite the extracted text into a conversational podcast script.
• Audio Generation: Converts the podcast script into audio using the SpeechT5 TTS model.

Table of Contents

• Installation
• Usage
• Directory Structure
• Dependencies
• License

Installation

To use the Audio Summarizer package, ensure you have Python 3.7+ and pip installed. Then, install the required dependencies:

pip install -r requirements.txt

Usage

To generate an audio podcast from a PDF document, run the following command:

python completed.py <pdf_path> [options]

Arguments

• pdf_path: The path to the PDF file you want to convert to audio.
• --modelOllama: Specify the Ollama model to use for text processing (default: llama3.2).
• --modelHugging: Specify the Hugging Face model for text generation (default: facebook/opt-125m).
• --max_tokens: Set the maximum number of tokens for the generation (default: 8126).
• --temperature: Set the sampling temperature for text generation (default: 1.0).
• --pdf_text_file: The filename to save extracted PDF text (default: pdf_text.txt).
• --cleaned_text_file: The filename to save cleaned text (default: cleaned.txt).
• --audiomodelname: The TTS model to use for generating the audio (default: suno/bark-small).

Example

python completed.py my_document.pdf --modelOllama llama3.2 --modelHugging facebook/opt-125m --audiomodelname suno/bark-small

This will process the PDF my_document.pdf, clean and rewrite the text, and generate audio output in the output_audio directory.

Directory Structure

./
├── AudioGenerator.py      # Handles audio generation using TTS models
├── completed.py           # Main script for PDF processing and audio generation
├── FileManager.py         # Manages file reading, cleaning, and GPU memory management
└── QueryHandler.py        # Handles text processing, chunking, and interaction with Hugging Face and Ollama models

Detailed Workflow

1. PDF Pre-Processing: Extract text from the PDF using the FileManager.read_pdf method.
2. Text Cleaning and Script Generation: The TextProcessor.clean_text method cleans the extracted text and generates a conversational podcast script using Hugging Face and Ollama models.
3. Audio Generation: The AudioGenerator.generate_audio_from_text method splits the text into chunks and generates audio using the SpeechT5 TTS model.

Dependencies

• torch: Required for deep learning models and TTS pipeline.
• transformers: Hugging Face library for using pre-trained models.
• datasets: Hugging Face dataset library for loading speaker embeddings.
• soundfile: For saving the generated audio to files.
• pydub: Used for audio manipulation (if necessary).
• pdfplumber: Used to extract text from PDF files.

Install Dependencies

You can install the dependencies with the following command:

pip install -r requirements.txt

License

This repository is licensed under the MIT License. See the LICENSE file for more details.

---

Feel free to customize the README further if needed!