Update app.py

app.py

@@ -1,86 +1,63 @@
 # import part
 import streamlit as st
 from transformers import pipeline
-import os
+import torch
 
 # function part
 # img2text
 def img2text(url):
-    try:
-        image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
-        text = image_to_text_model(url)[0]["generated_text"]
-        # Remove unwanted words like "illustration"
-        text = text.replace("illustration", "").strip()
-        # Make the caption more fun and happy
-        fun_caption = f"🌟 Wow! This picture shows {text.lower()}. Let’s turn it into a fun story! 🌟"
-        return fun_caption
-    except Exception as e:
-        st.error(f"Oops! Something went wrong while looking at your picture. Please try again! 😊")
-        return None
+    image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
+    text = image_to_text_model(url)[0]["generated_text"]
+    # Make the caption simple and fun for kids
+    fun_caption = f"Look what we found! 🎨 {text}"
+    return fun_caption
 
 # text2story
 def text2story(text):
-    try:
-        # Use a better model for text generation (e.g., GPT-Neo)
-        story_generator = pipeline("text-generation", model="EleutherAI/gpt-neo-125M")
-        # Add a fun and happy prompt to guide the story generation
-        prompt = f"One sunny day, {text}. "
-        story = story_generator(prompt, max_length=100, num_return_sequences=1)[0]["generated_text"]
-        # Remove any unwanted text (e.g., usernames, special characters)
-        story = " ".join([word for word in story.split() if not word.startswith("@") and not word.startswith("http")])
-        # Make the story more fun by adding a happy ending
-        happy_story = story + " And everyone had a big smile on their faces at the end of the day! 😄🌈"
-        return happy_story
-    except Exception as e:
-        st.error(f"Oops! Something went wrong while creating your story. Please try again! 😊")
-        return None
+    story_generator = pipeline("text-generation", model="distilgpt2")
+    # Generate a story with a maximum of 90 words
+    story = story_generator(text, max_length=90, num_return_sequences=1)[0]["generated_text"]
+    # Ensure the story does not exceed 90 words
+    story = " ".join(story.split()[:90])  # Truncate to 90 words
+    # Make the story simple and fun for kids
+    fun_story = f"Once upon a time... 🌟 {story}"
+    return fun_story
 
 # text2audio
 def text2audio(story_text):
-    try:
-        # Use a reliable TTS model (e.g., ESPnet's VITS model)
-        tts_pipeline = pipeline("text-to-speech", model="espnet/kan-bayashi_ljspeech_vits")
-        audio_output = tts_pipeline(story_text)
-        audio_file = "story_audio.wav"
-        # Save the audio file
-        with open(audio_file, "wb") as f:
-            f.write(audio_output["audio"])
-        return audio_file
-    except Exception as e:
-        st.error(f"Oops! Something went wrong while turning your story into audio. Please try again! 😊")
-        return None
+    tts_pipeline = pipeline("text-to-speech", model="espnet/kan-bayashi_ljspeech_vits")
+    audio_data = tts_pipeline(story_text)
+    return audio_data
 
 # main part
-st.set_page_config(page_title="Story Maker", page_icon="😊")
-st.header("😄 Story Maker: Turn Your Picture into a Happy Story! 😄")
-uploaded_file = st.file_uploader("📷 Choose a picture to create a fun story...", type=["jpg", "jpeg", "png"])
+st.set_page_config(page_title="Story Maker", page_icon="🦜")
+st.header("Story Maker: Turn Your Picture into a 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="Your fun picture!", use_container_width=True)
+    st.image(uploaded_file, caption="Your Picture", use_container_width=True)
 
     # Stage 1: Image to Text
-    st.write("🌟 Let’s see what’s in your picture... 🌟")
+    st.text('✨ Discovering what’s in your picture...')
     scenario = img2text(uploaded_file.name)
-    if scenario:
-        st.write(scenario)
-
-        # Stage 2: Text to Story
-        st.write("📖 Turning your picture into a fun story... 📖")
-        story = text2story(scenario)
-        if story:
-            st.write("📖 Here’s your fun story:")
-            st.write(story)
-
-            # Stage 3: Story to Audio data
-            st.write("🎤 Getting ready to tell your story... 🎤")
-            audio_file = text2audio(story)
-            if audio_file:
-                # Play button
-                if st.button("🎧 Listen to Your Story!"):
-                    st.audio(audio_file, format="audio/wav")
-                # Clean up the audio file after playing
-                os.remove(audio_file)
+    st.write(f"Here’s what we found: {scenario}")
+
+    # Stage 2: Text to Story
+    st.text('🎭 Creating a fun story for you...')
+    story = text2story(scenario)
+    st.write(story)
+
+    # Stage 3: Story to Audio data
+    st.text('🔊 Turning your story into audio...')
+    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'])