app.py

import streamlit as st
from transformers import pipeline
from huggingface_hub import cached_download
import torch
from PIL import Image

# Pre-trained models
text_generator = pipeline("text-generation", model="gpt2")
question_answering = pipeline("question-answering", model="distilbert-base-cased-squad2")
image_generator = pipeline("image-generation", model="sd-v1-diffusion")

# Download models if not already present
models = ["gpt2", "distilbert-base-cased-squad2", "sd-v1-diffusion"]
for model in models:
    cached_download(repo_id=f"huggingface/transformers/{model}", force_download=True)

# Story prompt input
prompt = st.text_input("Start your story with...", value="Once upon a time...")

# Generate story
generated_text = text_generator(prompt, max_length=1024)[0]["generated_text"]

# Text-to-speech and Text-to-text buttons
st.button("Speak the story", on_click=lambda: speak(generated_text))
generated_text_2 = st.button("Generate different story", on_click=lambda: generate_text(prompt))

# Generate question from story
question = st.text_input("Ask a question about the story...")
answer = question_answering(question=question, context=generated_text)["answer"]

# Generate image
image = image_generator(prompt="Image related to the story:")
image = Image.open(torch.ByteStorage.from_buffer(image[0]["image"]).read())

# Streamlit layout
st.title("Your Story")
st.write(generated_text)
st.image(image)

# Question and answer section
st.subheader("Ask and Learn")
st.write(f"Question: {question}")
st.write(f"Answer: {answer}")

# Text-to-text functionality
if generated_text_2:
    generated_text_2 = text_generator(prompt, max_length=1024)[0]["generated_text"]
    st.write("**New Story:**")
    st.write(generated_text_2)

# Text-to-speech functionality
def speak(text):
    # Your preferred text-to-speech implementation
    # (e.g., ResponsiveVoice, gTTS) goes here
    pass

st.write("* This app is still under development and may not always generate accurate or coherent results.")
st.write("* Please be mindful of the content generated by the AI models.")