Update app.py

app.py

@@ -46,7 +46,7 @@ def find_mp4_files(directory):
     for root, dirs, files in os.walk(directory):
         for file in files:
             # Check if the file ends with .mp4
-            if file.endswith(".wav"):
+            if file.endswith(".mp4") and file[:3] == 'est':
                 mp4_files.append(os.path.join(root, file))
     
     return mp4_files
@@ -61,7 +61,7 @@ def fn_clearvoice_tse(input_video):
     output_list = find_mp4_files('path_to_output_videos_tse/')
     print(output_list)
     
-    return output_list[0], output_list[1]
+    return output_list
 
 demo = gr.Blocks()
 
@@ -117,13 +117,12 @@ tse_demo = gr.Interface(
         gr.Video(label="Input Video"),
     ],
     outputs = [
-        gr.Audio(label="Output Audio", type="filepath"),
-        gr.Audio(label="Output Audio", type="filepath"),
+        gr.Video(label="Output Video List", type="filepath", multiple=True)
     ],
-    title = "ClearVoice: Speech Separation",
-    description = ("Gradio demo for Speech separation with ClearVoice. The model (MossFormer2 backbone) supports 2 speakers' audio mixtures with 16 kHz sampling rate. "
-                   "We provide the generalized models trained on large scale of data for handling independent speakers and various of background environments. "
-                    "To test it, simply upload your audio, or click one of the examples to load them. Read more at the links below."),
+    title = "ClearVoice: Audio-visual speaker extraction",
+    description = ("Gradio demo for audio-visual speaker extraction with ClearVoice. The model (AV_MossFormer2_TSE_16K) supports 16 kHz sampling rate. "
+                   "We provide the generalized models trained on mid-scale of data for handling independent speakers and various of background environments. "
+                    "To test it, simply upload your video, or click one of the examples to load them. Read more at the links below."),
     article = ("<p style='text-align: center'><a href='https://arxiv.org/abs/2302.11824' target='_blank'>MossFormer: Pushing the Performance Limit of Monaural Speech Separation using Gated Single-Head Transformer with Convolution-Augmented Joint Self-Attentions</a> | <a href='https://github.com/alibabasglab/MossFormer' target='_blank'>Github Repo</a></p>"
               "<p style='text-align: center'><a href='https://arxiv.org/abs/2312.11825' target='_blank'>MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation</a> | <a href='https://github.com/alibabasglab/MossFormer2' target='_blank'>Github Repo</a></p>"),
     examples = [