app.py

import streamlit as st
from transformers import pipeline
import soundfile as sf
import io

# --- Load models once at the beginning ---
image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
text_to_story_model = pipeline("text-generation", model="pranavpsv/genre-story-generator-v2")
story_to_audio_model = pipeline("text-to-speech", model="facebook/mms-tts-eng")

# --- Define Functions ---
# img2text
def img2text(image_path):
    text = image_to_text_model(image_path)[0]["generated_text"]
    return text

# text2story
def text2story(text):
    story_text = text_to_story_model(text, max_new_tokens=150)[0]['generated_text']
    words = story_text.split()
    if len(words) > 100:
        story_text = ' '.join(words[:100]) + '.'
    return story_text

# text2audio
def text2audio(story_text):
    speech_data = story_to_audio_model(story_text)
    return speech_data

# play_audio
def play_audio(audio_data):
    audio_buffer = io.BytesIO()
    sf.write(audio_buffer, audio_data['audio'], audio_data['sampling_rate'], format='WAV')
    audio_buffer.seek(0)
    st.audio(audio_buffer, format='audio/wav')

# --- Streamlit App ---
st.set_page_config(page_title="Your Image to Audio Story", page_icon="🦜")
st.header("Turn Your Image to Audio Story")

uploaded_file = st.file_uploader("Select an Image...")

if uploaded_file is not None:
    bytes_data = uploaded_file.getvalue()
    with open(uploaded_file.name, "wb") as file:
        file.write(bytes_data)
    st.image(uploaded_file, caption="Uploaded Image", use_container_width=True)

    # Stage 1: Image to Text
    if "scenario" not in st.session_state:
        with st.spinner('Processing image to text...'):
            st.session_state.scenario = img2text(uploaded_file.name)
    st.subheader("Image Description:")
    st.write(st.session_state.scenario)

    # Stage 2: Text to Story
    if "story" not in st.session_state:
        with st.spinner('Generating a story...'):
            st.session_state.story = text2story(st.session_state.scenario)
    st.subheader("Generated Story:")
    st.write(st.session_state.story)

    # Stage 3: Story to Audio
    if "audio_data" not in st.session_state:
        with st.spinner('Generating audio narration...'):
            st.session_state.audio_data = text2audio(st.session_state.story)

    # Play Audio Button
    if st.button("Play Audio"):
        if "audio_data" in st.session_state:
            play_audio(st.session_state.audio_data)
        else:
            st.warning("Please generate the audio first.")