Update app.py

app.py

@@ -5,20 +5,18 @@ import numpy as np
 import torch
 import spaces
 
-# Load the model
+# load the model
 print("Loading model...")
 model_id = "badrex/mms-300m-arabic-dialect-identifier"
 classifier = pipeline("audio-classification", model=model_id, device='cuda')
 print("Model loaded successfully")
 print("Model moved to GPU successfully")
 
-
 @spaces.GPU
 def predict(audio_segment, sr=16000):
     return classifier({"sampling_rate": sr, "raw": audio_segment})
 
-
-# Define dialect mapping
+# define dialect mapping
 dialect_mapping = {
     "MSA": "Modern Standard Arabic (MSA) - العربية الفصحى الحديثة",
     "Egyptian": "Egyptian Arabic -  اللهجة المصرية العامية",
@@ -31,16 +29,12 @@ def predict_dialect(audio):
     if audio is None:
         return {"Error": 1.0}
     
-    # The audio input from Gradio is a tuple of (sample_rate, audio_array)
     sr, audio_array = audio
     
-    # Process the audio input
     if len(audio_array.shape) > 1:
-        audio_array = audio_array.mean(axis=1)  # Convert stereo to mono
+        audio_array = audio_array.mean(axis=1)
 
-    # Convert audio to float32 if it's not already (fix for Chrome recording issue)
     if audio_array.dtype != np.float32:
-        # Normalize to [-1, 1] range as expected by the model
         if audio_array.dtype == np.int16:
             audio_array = audio_array.astype(np.float32) / 32768.0
         else:
@@ -48,10 +42,8 @@ def predict_dialect(audio):
     
     print(f"Processing audio: sample rate={sr}, shape={audio_array.shape}")
     
-    # Classify the dialect
     predictions = predict(sr=sr, audio_segment=audio_array)
     
-    # Format results for display
     results = {}
     for pred in predictions:
         dialect_name = dialect_mapping.get(pred['label'], pred['label'])
@@ -59,110 +51,36 @@ def predict_dialect(audio):
     
     return results
 
-# Manually prepare example file paths without metadata
+# prepare examples
 examples = []
 examples_dir = "examples"
 if os.path.exists(examples_dir):
     for filename in os.listdir(examples_dir):
         if filename.endswith((".wav", ".mp3", ".ogg")):
             examples.append([os.path.join(examples_dir, filename)])
-    
     print(f"Found {len(examples)} example files")
 else:
     print("Examples directory not found")
 
+# clean description without problematic HTML
+description = """
+By <a href="https://badrex.github.io/">Badr Alabsi</a> with ❤️🤍💚
 
+This is a demo for the accurate and robust Transformer-based <a href="https://huggingface.co/badrex/mms-300m-arabic-dialect-identifier">model</a> for Spoken Arabic Dialect Identification (ADI). 
+From just a short audio clip (5-10 seconds), the model can identify Modern Standard Arabic (MSA) as well as four major regional Arabic varieties: Egyptian Arabic, Peninsular Arabic (Gulf, Yemeni, and Iraqi), Levantine Arabic, and Maghrebi Arabic.
 
-# Custom CSS for better styling
-custom_css = """
-<style>
-.centered-content {
-    text-align: center;
-    max-width: 800px;
-    margin: 0 auto;
-    padding: 20px;
-}
-
-.logo-image {
-    width: 200px;
-    height: auto;
-    margin: 20px auto;
-    display: block;
-}
-
-.description-text {
-    font-size: 16px;
-    line-height: 1.6;
-    margin-bottom: 20px;
-}
-
-.dialect-list {
-    font-size: 15px;
-    line-height: 1.8;
-    text-align: left;
-    max-width: 600px;
-    margin: 0 auto;
-}
-
-.highlight-text {
-    font-size: 16px;
-    color: #2563eb;
-    margin: 20px 0;
-}
-
-.footer-text {
-    font-size: 13px;
-    color: #6b7280;
-    margin-top: 20px;
-}
-</style>
-"""
-
-"""
-<p style="font-size: 15px; line-height: 1.8;">
-                    <strong>The following Arabic language varieties are supported:</strong>
-                    <br><br>
-                    ✦ <strong>Modern Standard Arabic (MSA)</strong> - The formal language of media and education
-                    <br>
-                    ✦ <strong>Egyptian Arabic</strong> - The dialect of Cairo, Alexandria, and popular Arabic cinema
-                    <br>
-                    ✦ <strong>peninsular Arabic</strong> - Spoken across Saudi Arabia, UAE, Kuwait, Qatar, Bahrain, and Oman
-                    <br>
-                    ✦ <strong>Levantine Arabic</strong> - The dialect of Syria, Lebanon, Jordan, and Palestine
-                    <br>
-                    ✦ <strong>Maghrebi Arabic</strong> - The distinctive varieties of Morocco, Algeria, Tunisia, and Libya
-                    </p>
-                    <br>
+Simply **upload an audio file** 📤 or **record yourself speaking** 🎙️⏺️ to try out the model!
 """
 
-# Create the Gradio interface
 demo = gr.Interface(
     fn=predict_dialect,
     inputs=gr.Audio(),
     outputs=gr.Label(num_top_classes=5, label="Predicted Dialect"),
     title="Tamyïz 🍉 Arabic Dialect Identification in Speech",
-    description="""
-        <div class="centered-content">
-            <div>
-                <p>
-                By <a href="https://badrex.github.io/" style="color: #2563eb;">Badr Alabsi</a> with ❤��🤍💚 
-                </p>
-                <br>
-                <p style="font-size: 15px; line-height: 1.8;">
-                This is a demo for the accurate and robust Transformer-based <a href="https://huggingface.co/badrex/mms-300m-arabic-dialect-identifier" style="color: #FF5349;">model</a> for Spoken Arabic Dialect Identification (ADI). 
-                From just a short audio clip (5-10 seconds), the model can identify Modern Standard Arabic (<strong>MSA</strong>) as well as four major regional Arabic varieties: <strong>Egyptian</strong> Arabic, <strong>Peninsular</strong> Arabic (Gulf, Yemeni, and Iraqi), <strong>Levantine</strong> Arabic, and <strong>Maghrebi</strong> Arabic. 
-                <br>                   
-                <p style="font-size: 15px; line-height: 1.8;">
-                Simply <strong>upload an audio file</strong> 📤 or <strong>record yourself speaking</strong> 🎙️⏺️ to try out the model!
-                </p>
-            </div>
-        </div>
-        """,
+    description=description,
     examples=examples if examples else None,
-    cache_examples=False,  # Disable caching to avoid issues
-    #theme=gr.themes.Default(font=[gr.themes.GoogleFont("Amiri"), "Arial", "serif"]),
+    cache_examples=False,
     flagging_mode=None
 )
 
-# Launch the app
 demo.launch(share=True)