src/streamlit_app.py

import altair as alt
import numpy as np
import pandas as pd
import streamlit as st
from transformers import pipeline
from PIL import Image

# Creates a brief description for the pictures
def generate_caption(image):
    with st.spinner("Analysing the Pictures for Key Message..."):
        # Loads the BLIP model to examine and describe the picture
        image_to_text = pipeline("image-to-text", model="Salesforce/blip-image-captioning-large")
        caption = image_to_text(image)[0]["generated_text"]
    return caption

# Builds a story from the picture’s description
def generate_story(caption):
    with st.spinner("Enhancing the Story for better Details..."):
        # Uses the text generation model to create a story based on the description
        pipe = pipeline("text-generation", model="TheBloke/phi-2-GGUF")
        story = pipe(caption)[0]['generated_text']
    return story

# Turns the story into audio
def generate_audio(story):
    with st.spinner("Turning story into News audio..."):
        # Uses a speech model to turn description into audio
        pipe = pipeline("text-to-speech", model="hexgrad/Kokoro-82M")
        audio = pipe(story)
    return audio

# Streamlit UI: Makes a simple interface to generate the audio

# Displays the title
st.title("Tool for the Reporter - Turning the News Photo into Audio")

# Describes the app for users
st.write("Please upload the News Photo within 200MB")

# Allows picture uploads
uploaded_file = st.file_uploader("Upload the Photo below", type=["png", "jpg", "jpeg"])

if uploaded_file is not None:
    # Shows the uploaded picture
    image = Image.open(uploaded_file)
    st.image(image, caption="Your Picture!", use_container_width=True)

    # Gets the picture’s description
    image_caption = generate_caption(image)
    st.subheader("Phot Description:")
    st.write(image_caption)

    # Generate the News descriptions
    story_telling = generate_story(image_caption)
    st.subheader("The News:")
    st.write(story_telling)

    # Generates audio
    audio = generate_audio(story_telling)
    if st.button("Hear the News"):
        st.audio(audio['audio'],
                 format="audio/wav",
                 start_time=0,
                 sample_rate=audio['sampling_rate'])