app.py

import subprocess

subprocess.run(["pip", "install", "PyPDF2", "transformers", "bark", "gradio"])

import PyPDF2
from transformers import pipeline
from bark import SAMPLE_RATE, generate_audio, preload_models
import gradio as gr

def summarize_and_convert_to_audio(pdf_file, abstract_page):
    # Save the uploaded file
    pdf_path = "/tmp/uploaded_file.pdf"
    pdf_file.save(pdf_path)

    # Convert abstract_page to integer
    abstract_page = int(abstract_page)

    with open(pdf_path, 'rb') as file:
        pdf_reader = PyPDF2.PdfReader(file)
        
        # Get the abstract page text
        abstract_page_text = pdf_reader.pages[abstract_page - 1].extract_text()

    summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
    summary = summarizer(abstract_page_text, max_length=20, min_length=20)

    preload_models()

    text = summary[0]['summary_text']
    audio_array = generate_audio(text)

    return Audio(audio_array, rate=SAMPLE_RATE)

iface = gr.Interface(
    fn=summarize_and_convert_to_audio,
    inputs=[
        gr.File("application/pdf", label="Upload PDF file"),
        "text"
    ],
    outputs="audio",
    live=True
)

iface.launch()