Update app.py

app.py

@@ -4,80 +4,54 @@ from PIL import Image
 import io
 from gtts import gTTS
 
-# Page config
 st.title("🖼️ → 📖 Image-to-Story Demo")
 st.write("Upload an image and watch as it’s captioned, turned into a short story, and even read aloud!")
 
-# Load and cache pipelines
 @st.cache_resource
 def load_captioner():
     return pipeline("image-to-text", model="unography/blip-large-long-cap")
 
 @st.cache_resource
 def load_story_gen():
-    return pipeline(
-        "text-generation",
-        model="gpt2",
-        tokenizer="gpt2"
-    )
+    return pipeline("text-generation", model="gpt2", tokenizer="gpt2")
 
 captioner = load_captioner()
 story_gen = load_story_gen()
 
-# 1) Image upload
-uploaded = st.file_uploader("Upload an image", type=["png", "jpg", "jpeg"])
+# 1) Upload (key='image' gives us st.session_state.image)
+uploaded = st.file_uploader("Upload an image", type=["png","jpg","jpeg"], key="image")
 if uploaded:
     img = Image.open(uploaded)
     st.image(img, use_column_width=True)
 
-    # 2) Generate caption
-    with st.spinner("Generating caption…"):
-        caps = captioner(img)
-        # `caps` is a list of dicts like [{"generated_text": "..."}]
-        caption = caps[0]["generated_text"]
-    st.write("**Caption:**", caption)
-
-    # 3) Generate story from caption
-    with st.spinner("Spinning up a story…"):
-        story_out = story_gen(
-            caption,
-            max_length=200,
-            num_return_sequences=1,
-            do_sample=True,
-            top_p=0.9
-        )
-        story = story_out[0]["generated_text"]
-    st.write("**Story:**", story)
-
-    # 4) Play story as audio
-    if st.button("🔊 Play Story Audio"):
+    # 2) Caption (once per upload)
+    if "caption" not in st.session_state:
+        with st.spinner("Generating caption…"):
+            st.session_state.caption = captioner(img)[0]["generated_text"]
+    st.write("**Caption:**", st.session_state.caption)
+
+    # 3) Story (once per upload)
+    if "story" not in st.session_state:
+        with st.spinner("Spinning up a story…"):
+            out = story_gen(
+                st.session_state.caption,
+                max_length=200,
+                num_return_sequences=1,
+                do_sample=True,
+                top_p=0.9
+            )
+            st.session_state.story = out[0]["generated_text"]
+    st.write("**Story:**", st.session_state.story)
+
+    # 4) Pre-generate audio buffer (once per upload)
+    if "audio_buffer" not in st.session_state:
         with st.spinner("Generating audio…"):
-            tts = gTTS(text=story, lang="en")
+            tts = gTTS(text=st.session_state.story, lang="en")
             buf = io.BytesIO()
             tts.write_to_fp(buf)
             buf.seek(0)
-        st.audio(buf.read(), format="audio/mp3")
-
-
-"""
-import streamlit as st
-from transformers import pipeline
-
-def main():
-    sentiment_pipeline = pipeline(model="distilbert/distilbert-base-uncased-finetuned-sst-2-english")
-
-    st.title("Sentiment Analysis with HuggingFace Spaces")
-    st.write("Enter a sentence to analyze its sentiment:")
+            st.session_state.audio_buffer = buf.read()
 
-    user_input = st.text_input("")
-    if user_input:
-        result = sentiment_pipeline(user_input)
-        sentiment = result[0]["label"]
-        confidence = result[0]["score"]
-
-        st.write(f"Sentiment: {sentiment}")
-        st.write(f"Confidence: {confidence:.2f}")
-
-if __name__ == "__main__":
-    main()
-"""
+    # 5) Play on demand
+    if st.button("🔊 Play Story Audio"):
+        st.audio(st.session_state.audio_buffer, format="audio/mp3")