app.py

# import part
import streamlit as st
from transformers import pipeline
import soundfile as sf
import numpy as np
import tempfile

# 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):
    story_text_model = pipeline("text-generation", model="google/gemma-2-9b-it")
    story = story_text_model(text, max_length=150)[0]['generated_text']
    return story

# text2audio
def text2audio(story_text):
    tts_model = pipeline("text-to-speech", model="tts_models/en/ljspeech/tacotron2")
    audio_data = tts_model(story_text)
    
    # Save audio to a temporary file
    audio_filename = tempfile.mktemp(suffix=".wav")
    sf.write(audio_filename, audio_data['audio'], audio_data['sampling_rate'])
    
    return audio_filename

# main part
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_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...')
    story = text2story(scenario)
    st.write(story)

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

    # Play button
    if st.button("Play Audio"):
        st.audio(audio_filename, format="audio/wav")