Update app.py

app.py

@@ -10,72 +10,71 @@ import tempfile
 st.set_page_config(page_title="Storyteller for Kids", layout="centered")
 st.title("🖼️ ➡️ 📖 Interactive Storyteller")
 
-# —––––––– Load & warm pipelines
+# —––––––– Load and warm pipelines
 @st.cache_resource
 def load_pipelines():
-    # 1) BLIP-base for captions
+    # BLIP-base for captions
     captioner = pipeline(
         "image-to-text",
         model="Salesforce/blip-image-captioning-base",
-        device=-1  # or -1 if you only have CPU
+        device=-1  # GPU if available, else -1
     )
-    # 2) Flan-T5-Large for instruction-driven stories
+    # Flan-T5-Large for stories
     storyteller = pipeline(
         "text2text-generation",
         model="google/flan-t5-large",
         device=0
     )
-
-    # Warm-up to avoid first-call lag
+    # Warm-up runs so user-facing calls are fast
     dummy = Image.new("RGB", (384, 384), color=(128, 128, 128))
     captioner(dummy)
     storyteller("Warm up", max_new_tokens=1)
-
     return captioner, storyteller
 
 captioner, storyteller = load_pipelines()
 
 # —––––––– Main UI
-uploaded = st.file_uploader("Upload an image:", type=["jpg","jpeg","png"])
+uploaded = st.file_uploader("Upload an image:", type=["jpg", "jpeg", "png"])
 if uploaded:
-    # 1) Preprocess & display
+    # 1) Preprocess image
     image = Image.open(uploaded).convert("RGB")
     image = image.resize((384, 384), Image.LANCZOS)
     st.image(image, caption="Your image", use_container_width=True)
 
     # 2) Caption
-    with st.spinner("🔍 Generating caption..."):
+    with st.spinner("🔍 Generating caption…"):
         cap = captioner(image)[0]["generated_text"].strip()
     st.markdown(f"**Caption:** {cap}")
 
-    # 3) Story
+    # 3) Build a dynamic prompt
     prompt = (
-        f"Image description: “{cap}”.\n"
+        f"Here is an image description: “{cap}”.\n"
         "Write an 80–100 word playful story for 3–10 year-old children that:\n"
-        "1) Sets the scene around the panda.\n"
-        "2) Describes what it’s doing and how it feels.\n"
-        "3) Ends with a fun conclusion.\n\n"
+        "1) Describes the scene and subject from the description.\n"
+        "2) Explains what the subject is doing and how it feels.\n"
+        "3) Concludes with a fun, imaginative ending.\n\n"
         "Story:"
     )
-    with st.spinner("✍️ Generating story..."):
-        out = storyteller(
+
+    # 4) Generate the story
+    with st.spinner("✍️ Writing the story…"):
+        output = storyteller(
             prompt,
-            max_new_tokens=130,
+            max_new_tokens=120,
             do_sample=True,
             temperature=0.7,
             top_p=0.9,
             top_k=50,
-            repetition_penalty=1.3,
+            repetition_penalty=1.2,
             no_repeat_ngram_size=3
         )
-        # text2text pipeline returns only the generated part
-        story = out[0]["generated_text"].strip()
+        story = output[0]["generated_text"].strip()
 
     st.markdown("**Story:**")
     st.write(story)
 
-    # 4) Text-to-Speech via gTTS
-    with st.spinner("🔊 Converting to speech..."):
+    # 5) Text-to-Speech
+    with st.spinner("🔊 Converting to speech…"):
         tts = gTTS(text=story, lang="en")
         tmp = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False)
         tts.write_to_fp(tmp)