app.py

import streamlit as st
from transformers import pipeline

# function part
# img2text
def img2text(url):
    image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
    text = image_to_text_model(url)[0]["generated_text"]
    return text

# text2story
def text2story(text):
    #messages = f"""Generate a vivid story in 100-120 words using the following sentences: "{text}".
    #Structure requirements:
    #1. Only 1 paragraphs and the story mest have beginning, middle, end
    #2. Include at least two named characters
    #3. Use descriptive language
    
    #Story:"""
    #messages = f"""Generate a story of about 100 words using the following sentences: {text}."""
    messages = [
        {"role": "system", "content": "You are Qwen, created by Alibaba Cloud. You are a helpful assistant."},
        {"role": "user", "content": f"""Please help generate a vivid story in 200 words using the following sentences: "{text}".The story must contain:Beginning, middle, end and use descriptive language"""},
    ]
    text_to_story_model= pipeline("text-generation", model="Qwen/Qwen2.5-0.5B-Instruct")
    story_text = text_to_story_model(
        messages,
        max_length=400,  
        min_length=100,
        do_sample=True,
        temperature=0.65
    )[0]['generated_text'][2]['content']
    return story_text
    
# text2audio
def text2audio(story_text):
    text_to_audio_model = pipeline("text-to-speech", model="facebook/mms-tts-eng")
    #audio_data = ""     # to be completed
    audio_data = text_to_audio_model(story_text)
    return audio_data

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:
    print(uploaded_file)
    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_column_width=True)

    #Stage 1: Image to Text
    st.text('Processing img2text...')
    scenario = img2text(uploaded_file.name)
    st.write(scenario)

    #Stage 2: Text to Story
    st.text('Generating a story...')
    #scenario = "children playing in the park illustration"
    story = text2story(scenario)
    st.write(story)

    #Stage 3: Story to Audio data
    st.text('Generating audio data...')
    audio_data =text2audio(story)

    # Play button
    if st.button("Play Audio"):
        st.audio(audio_data['audio'],
                    format="audio/wav",
                    start_time=0,
                    sample_rate = audio_data['sampling_rate'])