Update app.py

app.py

@@ -1,65 +1,86 @@
 import streamlit as st
-from audiorecorder import audiorecorder
-from transformers import pipeline
-from diffusers import StableDiffusionPipeline, DDIMScheduler
+from diffusers import StableDiffusionPipeline
 import torch
-from tempfile import NamedTemporaryFile
+from PIL import Image, ImageDraw, ImageFont
+import tempfile
+import speech_recognition as sr
 
-st.set_page_config(page_title="AI Meme Generator", page_icon="🎭")
-st.title("🎭 AI Meme Generator (Voice + Text)")
-
-# Load Whisper
-@st.cache_resource
-def load_asr():
-    return pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
-
-# Load Stable Diffusion with safe scheduler
+# Load Stable Diffusion
 @st.cache_resource
-def load_sd():
-    device = "cuda" if torch.cuda.is_available() else "cpu"
-    pipe = StableDiffusionPipeline.from_pretrained(
-        "runwayml/stable-diffusion-v1-5",   # ✅ Hugging Face model URL
-        scheduler=DDIMScheduler.from_pretrained(
-            "runwayml/stable-diffusion-v1-5", subfolder="scheduler"
-        ),
-        torch_dtype=torch.float16 if device == "cuda" else torch.float32
-    ).to(device)
+def load_model():
+    model_id = "runwayml/stable-diffusion-v1-5"
+    pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+    pipe = pipe.to("cuda" if torch.cuda.is_available() else "cpu")
     return pipe
 
-asr = load_asr()
-sd_pipe = load_sd()
+sd_pipe = load_model()
+
+# Meme generator function
+def generate_meme(prompt, caption=""):
+    # Step 1: Generate base image
+    image = sd_pipe(prompt, num_inference_steps=30).images[0]
+
+    # Step 2: Add caption text
+    if caption:
+        draw = ImageDraw.Draw(image)
+        try:
+            font = ImageFont.truetype("DejaVuSans-Bold.ttf", 48)
+        except:
+            font = ImageFont.load_default()
+
+        W, H = image.size
+        text = caption.upper()
+
+        # Wrap long text
+        import textwrap
+        lines = textwrap.wrap(text, width=25)
+
+        y_text = 20
+        for line in lines:
+            w, h = draw.textsize(line, font=font)
+            draw.text(((W - w) / 2, y_text), line, font=font,
+                      fill="white", stroke_width=3, stroke_fill="black")
+            y_text += h + 10
+
+    return image
+
+# Speech-to-text
+def speech_to_text(audio_file):
+    recognizer = sr.Recognizer()
+    with sr.AudioFile(audio_file) as source:
+        audio = recognizer.record(source)
+    try:
+        return recognizer.recognize_google(audio)
+    except:
+        return "Could not recognize speech."
 
-# Generate meme (with safe inference steps)
-def generate_meme(prompt):
-    return sd_pipe(prompt, num_inference_steps=30).images[0]
+# Streamlit UI
+st.title("🎭 Meme Generator (Text & Voice)")
 
-# Tabs
-tab1, tab2 = st.tabs(["📝 Text to Meme", "🎤 Voice to Meme"])
+mode = st.radio("Choose Input Mode:", ["Text", "Voice"])
 
-# ---------------- Text-to-Meme ----------------
-with tab1:
-    text_input = st.text_area("Enter your meme idea")
-    if st.button("Generate Meme", key="text_meme"):
+if mode == "Text":
+    text_input = st.text_area("Enter meme text:")
+    if st.button("Generate Meme"):
         if text_input.strip():
-            with st.spinner("Generating meme..."):
-                img = generate_meme(f"Meme style funny cartoon with text: {text_input}")
-                st.image(img, caption="Generated Meme")
+            img = generate_meme("funny cartoon meme background", caption=text_input)
+            st.image(img, caption="Generated Meme", use_column_width=True)
         else:
-            st.warning("Please enter some text!")
+            st.warning("Please enter some text.")
 
-# ---------------- Voice-to-Meme ----------------
-with tab2:
-    st.write("🎤 Record your voice below and create a meme!")
-    audio = audiorecorder("Click to record", "Click to stop recording")
+else:
+    audio_file = st.file_uploader("Upload voice file (.wav)", type=["wav"])
+    if audio_file:
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
+            tmp_file.write(audio_file.read())
+            tmp_path = tmp_file.name
 
-    if len(audio) > 0:
-        # Save temp audio file
-        with NamedTemporaryFile(suffix=".wav", delete=False) as f:
-            audio.export(f.name, format="wav")
-            text = asr(f.name)["text"]
+        text = speech_to_text(tmp_path)
+        st.write(f"📝 Transcribed Text: {text}")
 
-        st.success(f"Recognized Text: {text}")
-        if st.button("Generate Meme from Voice", key="voice_meme"):
-            with st.spinner("Transcribing and generating meme..."):
-                img = generate_meme(f"Meme style funny cartoon with text: {text}")
-                st.image(img, caption="Generated Meme")
+        if st.button("Generate Meme from Voice"):
+            if text.strip():
+                img = generate_meme("funny cartoon meme background", caption=text)
+                st.image(img, caption="Generated Meme", use_column_width=True)
+            else:
+                st.warning("Speech not recognized.")