Spaces:
Sleeping
Sleeping
| # Import necessary libraries | |
| import streamlit as st | |
| from transformers import pipeline | |
| from gtts import gTTS | |
| import os | |
| # Function to convert image to text using Hugging Face's BLIP model | |
| def img2text(url): | |
| """ | |
| Converts an image to text using the Salesforce/blip-image-captioning-base model. | |
| Args: | |
| url (str): Path to the image file. | |
| Returns: | |
| str: Generated text caption from the image. | |
| """ | |
| 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 | |
| # Function to generate a kid-friendly story from the text caption | |
| def text2story(text): | |
| """ | |
| Generates a kid-friendly story from the text caption using the aspis/gpt2-genre-story-generation model. | |
| Args: | |
| text (str): Text caption generated from the image. | |
| Returns: | |
| str: Generated story suitable for kids aged 3-10, within 95 words. | |
| """ | |
| # Add a prompt to ensure the story is kid-friendly and happy | |
| prompt = f"Write a happy, fun and complete story for kids aged 3-10 based on the following scenario: {text}. Keep the story under 120 words." | |
| # Load the text generation model | |
| story_generator = pipeline("text-generation", model="aspis/gpt2-genre-story-generation") | |
| # Generate the story | |
| story = story_generator(prompt, max_length=150, num_return_sequences=1)[0]["generated_text"] | |
| # Remove the prompt from the generated story | |
| story = story.replace(prompt, "").strip() | |
| # Ensure the story is within 95 words | |
| story_words = story.split() | |
| if len(story_words) > 98: | |
| story = " ".join(story_words[:98]) | |
| return story | |
| # Function to convert text to audio using gTTS | |
| def text2audio(story_text): | |
| """ | |
| Converts the generated story text to audio using gTTS. | |
| Args: | |
| story_text (str): The generated story text. | |
| Returns: | |
| str: Path to the generated audio file. | |
| """ | |
| # Convert text to speech | |
| tts = gTTS(text=story_text, lang='en') | |
| audio_file = "story_audio.mp3" | |
| tts.save(audio_file) | |
| return audio_file | |
| # Main application | |
| st.set_page_config(page_title="Picture Stories 🎨📖", page_icon="🦄") | |
| st.title("Picture Stories 🎨📖") | |
| st.markdown("### Turn your pictures into fun stories and listen to them! 🎉") | |
| # Instructions for kids | |
| st.markdown(""" | |
| **How to use this app:** | |
| 1. **Upload a picture** of something fun, like your favorite toy, a park, or your pet. | |
| 2. Wait for the app to **create a story** from your picture. | |
| 3. **Listen to the story** by clicking the "Play Audio" button. | |
| 4. Enjoy your fun story! 🎧 | |
| """) | |
| # Upload image | |
| uploaded_file = st.file_uploader("📷 **Upload your picture here!**", type=["jpg", "jpeg", "png"]) | |
| if uploaded_file is not None: | |
| # Save the uploaded file | |
| bytes_data = uploaded_file.getvalue() | |
| with open(uploaded_file.name, "wb") as file: | |
| file.write(bytes_data) | |
| # Display the uploaded image | |
| st.image(uploaded_file, caption="Your awesome picture!", use_container_width=True) | |
| # Stage 1: Image to Text | |
| st.text('✨ Turning your picture into words...') | |
| scenario = img2text(uploaded_file.name) | |
| st.write("**What we see:**", scenario) | |
| # Stage 2: Text to Story | |
| st.text('📖 Creating a fun story for you...') | |
| story = text2story(scenario) | |
| st.write("**Your story:**", story) | |
| # Stage 3: Story to Audio | |
| st.text('🎧 Turning your story into audio...') | |
| audio_file = text2audio(story) | |
| # Play button for the generated audio | |
| if st.button("🎵 **Play Audio**"): | |
| st.audio(audio_file, format="audio/mp3") | |
| # Clean up the generated audio file | |
| os.remove(audio_file) |