Create app.py

app.py

@@ -0,0 +1,64 @@
+import streamlit as st
+from transformers import pipeline
+from PIL import Image
+import io
+from gtts import gTTS
+import tempfile
+
+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!")
+
+@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")
+
+captioner = load_captioner()
+story_gen = load_story_gen()
+
+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)
+
+    # Caption
+    if "caption" not in st.session_state:
+        with st.spinner("Generating caption…"):
+            caps = captioner(img)
+            st.session_state.caption = caps[0] if isinstance(caps, list) else caps
+    st.write("**Caption:**", st.session_state.caption)
+
+    # Story
+    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)
+
+    # Prepare audio bytes once
+    if "audio_bytes" not in st.session_state:
+        with st.spinner("Generating audio…"):
+            tts = gTTS(text=st.session_state.story, lang="en")
+            buf = io.BytesIO()
+            tts.write_to_fp(buf)
+            st.session_state.audio_bytes = buf.getvalue()
+
+    # Play button
+    if st.button("🔊 Play Story Audio"):
+        # Write to a temp file
+        tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
+        tmp.write(st.session_state.audio_bytes)
+        tmp.flush()
+        tmp_path = tmp.name
+        tmp.close()
+        # Stream it
+        st.audio(tmp_path, format="audio/mp3")