Update app.py

app.py

@@ -247,6 +247,12 @@ def transcribe_once(audio_path, language_choice, beam_size, temperature):
     transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
     return transcription.strip()
 
+def normalize_word(word):
+    """Remove punctuation and normalize word for comparison"""
+    import string
+    # Remove punctuation and whitespace
+    return word.strip().translate(str.maketrans('', '', string.punctuation)).lower()
+
 def create_tabular_feedback(intended, actual, lang_choice):
     """Create clean, readable tabular feedback without background colors"""
     
@@ -272,6 +278,18 @@ def create_tabular_feedback(intended, actual, lang_choice):
         <h3 style='color: #2c3e50; margin-bottom: 20px; text-align: center;'>📊 Pronunciation Analysis</h3>
     """
     
+    # Show simple transliteration of target sentence for easier reading
+    if lang_choice in ["Tamil", "Malayalam"]:
+        feedback_html += f"""
+        <div style='margin-bottom: 25px; padding: 15px; border: 2px solid #3498db; border-radius: 8px; background: #f8f9fa;'>
+            <h4 style='color: #3498db; margin-bottom: 10px;'>🎯 Target Sentence (How to Read)</h4>
+            <div style='font-size: 20px; font-family: monospace; color: #2c3e50; line-height: 1.4;'>
+                <strong>Original:</strong> {intended}<br>
+                <strong>Read as:</strong> <span style='color: #e67e22; font-weight: bold;'>{intended_roman}</span>
+            </div>
+        </div>
+        """
+    
     # Overview table - completely clean
     feedback_html += """
     <div style='margin-bottom: 25px;'>
@@ -316,8 +334,10 @@ def create_tabular_feedback(intended, actual, lang_choice):
             <tbody>
     """
     
-    # Compare words using difflib
-    sm = difflib.SequenceMatcher(None, intended_words, actual_words)
+    # Compare words using difflib with normalized comparison
+    normalized_intended = [normalize_word(w) for w in intended_words]
+    normalized_actual = [normalize_word(w) for w in actual_words]
+    sm = difflib.SequenceMatcher(None, normalized_intended, normalized_actual)
     word_index = 0
     
     for tag, i1, i2, j1, j2 in sm.get_opcodes():