Update app.py

app.py

@@ -177,17 +177,34 @@ def infer(video_in):
     final_merged_out = merge_audio_to_video(video_in, wave_files[0])
     return wave_files[0], 'spectrogram.png', final_merged_out
 
-with gr.Blocks() as demo:
+css = """
+#col-container{
+    max-width: 800px;
+}
+"""
+with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown("# Video-To-Audio")
-        with gr.Row():
-            with gr.Column():
-                video_in = gr.Video(label='Video IN')
-                submit_btn = gr.Button("Submit")
-            with gr.Column():
-                output_sound = gr.Audio(label="Audio OUT")
-                output_spectrogram = gr.Image(label='Spectrogram')
-                merged_out = gr.Video(label="Merged video + generated audio")
+        gr.Markdown("# Video-to-Audio Generation with Hidden Alignment")
+        gr.HTML("""
+        <div style="display:flex;column-gap:4px;">
+            <a href='https://sites.google.com/view/vta-ldm'>
+                <img src='https://img.shields.io/badge/Project-Page-Green'>
+            </a>
+            <a href='https://huggingface.co/papers/2407.07464'>
+                <img src='https://img.shields.io/badge/HF-Paper-red'>
+            </a>
+            <a href='https://github.com/ariesssxu/vta-ldm'>
+                <img src='https://img.shields.io/badge/GitHub-Code-Blue'>
+            </a>
+        </div>
+        """)
+        with gr.Column():
+            video_in = gr.Video(label='Video IN')
+            submit_btn = gr.Button("Submit")
+        
+            output_sound = gr.Audio(label="Audio OUT")
+            output_spectrogram = gr.Image(label='Spectrogram')
+            merged_out = gr.Video(label="Merged video + generated audio")
         
     submit_btn.click(
         fn = infer,