Spaces:
Sleeping
Sleeping
| # import part | |
| import streamlit as st | |
| from transformers import pipeline | |
| from PIL import Image | |
| import io | |
| import numpy as np | |
| import soundfile as sf # For handling audio file operations | |
| # function part | |
| def generate_image_caption(image): | |
| """Generates a caption for the given image using a pre-trained model. | |
| Args: | |
| image: PIL Image object | |
| Returns: | |
| str: Generated caption text | |
| """ | |
| # Initialize image-to-text pipeline with BLIP model | |
| img2caption = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base") | |
| result = img2caption(image) | |
| return result[0]['generated_text'] | |
| def text2story(text): | |
| """Generates a children's story from text input using story generation model. | |
| Args: | |
| text: Input text prompt | |
| Returns: | |
| str: Generated story text | |
| """ | |
| # Craft prompt with specific requirements for children's stories | |
| story_prompt = f"Create a funny 100-word story for 8-year-olds about: {text}. Include: " | |
| story_prompt += "1) A silly character 2) Magical object 3) Sound effects 4) Happy ending" | |
| # Initialize text generation pipeline | |
| pipe = pipeline("text-generation", model="pranavpsv/genre-story-generator-v2") | |
| # Generate story with controlled randomness parameters | |
| story_text = pipe( | |
| story_prompt, | |
| max_new_tokens=200, # Limit story length | |
| temperature=0.9, # Control randomness (higher = more creative) | |
| top_k=50 # Limit vocabulary choices | |
| )[0]['generated_text'] | |
| # Clean output by splitting at the required ending marker | |
| return story_text.split("Happy ending")[-1].strip() | |
| def story_to_speech(story_text): | |
| """Converts story text to audio using text-to-speech model. | |
| Args: | |
| story_text: Story text to convert | |
| Returns: | |
| BytesIO: Audio data in WAV format | |
| """ | |
| # Initialize Bark text-to-speech pipeline | |
| tts_pipe = pipeline("text-to-speech", model="suno/bark-small") | |
| # Generate audio array (numpy array of sound samples) | |
| audio_output = tts_pipe(story_text, max_length=400) # Limit text length for stability | |
| # Convert numpy array to playable audio bytes | |
| audio_bytes = io.BytesIO() | |
| sf.write( | |
| audio_bytes, | |
| audio_output["audio"], | |
| audio_output["sampling_rate"], | |
| format='WAV' | |
| ) | |
| audio_bytes.seek(0) # Reset pointer for Streamlit audio player | |
| return audio_bytes | |
| def main(): | |
| """Main function for Streamlit application workflow""" | |
| # Configure page header | |
| st.title("📖 Image Story Generator with Audio") | |
| st.write("Upload an image to get a magical story read aloud!") | |
| # Image upload widget (supports JPG/PNG) | |
| uploaded_image = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"]) | |
| if uploaded_image: | |
| # Process image | |
| image = Image.open(uploaded_image).convert("RGB") # Ensure RGB format | |
| st.image(image, use_column_width=True) # Display uploaded image | |
| # Image analysis section | |
| with st.spinner("✨ Analyzing image..."): | |
| caption = generate_image_caption(image) | |
| # Display image understanding | |
| st.subheader("Image Understanding") | |
| st.write(caption) | |
| # Story generation section | |
| with st.spinner("📖 Writing story..."): | |
| story = text2story(caption) | |
| # Display generated story | |
| st.subheader("Magical Story") | |
| st.write(story) | |
| # Audio generation section | |
| if st.button("🎧 Read Story Aloud"): | |
| with st.spinner("🔊 Generating audio..."): | |
| try: | |
| # Convert story to audio (trim to 400 characters for model stability) | |
| audio_bytes = story_to_speech(story[:400]) | |
| # Display audio player | |
| st.audio(audio_bytes, format="audio/wav") | |
| except Exception as e: | |
| st.error(f"Error generating audio: {str(e)}") | |
| if __name__ == "__main__": | |
| # Start the Streamlit application | |
| main() |