change delete mechanism

app.py

@@ -423,7 +423,7 @@ def process_audio_for_matching(audio_file):
             <h3 style="color: #111827; margin: 0;">Vocal Matching Results</h3>
             <p style="color: #6b7280; margin: 5px 0;">Found {len(matches)} similar vocals in Covers80 dataset</p>
             <p style="color: #2563eb; margin: 5px 0; font-size: 0.9em;">🎵 Listen to original songs and extracted segments</p>
-            <p style="color: #9ca3af; margin: 5px 0; font-size: 0.85em;">💡 Scores above 60 generally indicate meaningful similarity for me haha..</p>
+            <p style="color: #9ca3af; margin: 5px 0; font-size: 0.85em;">💡 Scores above 50 generally indicate meaningful similarity for me haha..</p>
         </div>
         {matches_html}
     </div>
@@ -474,7 +474,8 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft(), title="Music Plagiarism D
             Authors: Seonghyeon Go, Yumin Kim | MIPPIA Inc. | Submitted to ICASSP 2026
         </p>
         <hr style="border: none; border-top: 1px solid #e5e7eb; margin: 15px 0;">
-        <p><strong>Demo Version Notice:</strong> This demo differs from the paper version and focuses exclusively on vocal segment transcription.</p>
+        <p><strong>Demo Version Notice:</strong> This demo differs from the paper version and focuses exclusively on vocal.</p>
+                <p> Please use this demo for only understanding the concept of segment-level matching!</p>
         <p style="font-size: 0.9em; color: #6b7280; margin: 8px 0;">
             Structure analysis has been excluded for optimization. Results are derived from all downbeats, 
             so segment boundaries may not align perfectly with musical phrases.