README.md

metadata

title: Audio Abstract42
emoji: 😻
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.7.1
app_file: app.py
pinned: false

PDF Audio Summarizer

This application summarizes PDF documents and converts the summary to audio.

How it works

The core logic is in the audio_pdf function. It:

1. Extracts raw text from the uploaded PDF using PyPDF2
2. Summarizes the text using LED-Based Summarization Model from HuggingFace Transformers. This uses AutoTokenizer and AutoModelForSeq2SeqLM to load the model and generate a summary
3. Converts the text summary to an audio file using gTTS (Google Text-to-Speech)

The summary and audio file are returned and displayed in the Gradio web interface.

Interface

The interface is created using Gradio. The key components are:

• File input to upload a PDF
• Text output to display the text summary
• Audio output to play the audio file

The interface is launched via iface.launch()

Dependencies

• PyPDF2
• Transformers
• gTTS
• Gradio
• torch
• numpy
• scipy
• io

Additional dependencies:

• torch: For neural network computations in Transformers
• numpy: For numerical processing
• scipy: For scientific computing
• io: To buffer the audio data

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference