updated gif

app.py

@@ -58,7 +58,7 @@
 #     description=description,
 # ).launch()
 
-
+ 
 
 # import os
 # import gradio as gr
@@ -70,17 +70,15 @@
 # from audio_separator import Separator  
 
 # def inference(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, backing_vocals):
+#     # Initially, show the loading GIF
+#     loading_gif_path = "7RwF.gif"
+#     gr.Image(loading_gif_path,visible=True) 
+
 #     os.makedirs("out", exist_ok=True)
 #     audio_path = 'test.wav'
 #     write(audio_path, audio[0], audio[1])
 #     device = 'cuda' if torch.cuda.is_available() else 'cpu'
 #     print(f"Using device: {device}")
-#     if device=='cuda':
-#         use_cuda=True
-#         print(f"Using device: {device}")
-#     else:
-#         use_cuda=False
-#         print(f"Using device: {device}")
 
 #     try:
 #         command = f"python3 -m demucs.separate -n htdemucs_6s -d {device} {audio_path} -o out"
@@ -88,14 +86,14 @@
 #         print("Demucs script output:", process.stdout.decode())
 #     except subprocess.CalledProcessError as e:
 #         print("Error in Demucs script:", e.stderr.decode())
-#         return [gr.Audio(visible=False)] * 8
+#         return [gr.Audio(visible=False)] * 8 + [loading_gif_path]
 
 #     try:
-#         separator = Separator("./out/htdemucs_6s/test/vocals.wav", model_name='UVR_MDXNET_KARA_2', use_cuda=device==use_cuda, output_format='wav')
+#         separator = Separator("./out/htdemucs_6s/test/vocals.wav", model_name='UVR_MDXNET_KARA_2', use_cuda=device=='cuda', output_format='wav')
 #         primary_stem_path, secondary_stem_path = separator.separate()
 #     except Exception as e:
 #         print("Error in custom separation:", str(e))
-#         return [gr.Audio(visible=False)] * 8
+#         return [gr.Audio(visible=False)] * 8 + [loading_gif_path]
 
 #     stem_paths = {
 #         "vocals": "./out/htdemucs_6s/test/vocals.wav" if vocals else None,
@@ -108,7 +106,9 @@
 #         "backing_vocals": secondary_stem_path if backing_vocals else None
 #     }
 
-#     return [gr.Audio(stem_paths[stem], visible=bool(stem_paths[stem])) for stem in stem_paths]
+#     # Once processing is done, hide the GIF by returning a transparent image
+#     gr.Image(visible=False)
+#     return [gr.Audio(stem_paths[stem], visible=bool(stem_paths[stem])) for stem in stem_paths]  
 
 # # Define checkboxes for each stem
 # checkbox_labels = ["Full Vocals", "Bass", "Drums", "Other", "Piano", "Guitar", "Lead Vocals", "Backing Vocals"]
@@ -123,7 +123,6 @@
 #     [gr.Audio(label=label, visible=False) for label in checkbox_labels],
 #     title=title,
 #     description=description,
-    
 # )
 
 # iface.launch()
@@ -132,6 +131,8 @@
 
 
 
+
+
 import os
 import gradio as gr
 from scipy.io.wavfile import write
@@ -141,32 +142,37 @@ import torch
 # Assuming audio_separator is available in your environment
 from audio_separator import Separator  
 
-def inference(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, backing_vocals):
-    # Initially, show the loading GIF
-    loading_gif_path = "7RwF.gif"
-    gr.Image(loading_gif_path,visible=True) 
+# Global variable to track the display state of the GIF
+show_loading_gif = True
+
+def start_loading():
+    global show_loading_gif
+    show_loading_gif = True
+
+def stop_loading():
+    global show_loading_gif
+    show_loading_gif = False
 
+def process_audio(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, backing_vocals):
+    # Audio processing logic
     os.makedirs("out", exist_ok=True)
     audio_path = 'test.wav'
     write(audio_path, audio[0], audio[1])
     device = 'cuda' if torch.cuda.is_available() else 'cpu'
-    print(f"Using device: {device}")
 
     try:
         command = f"python3 -m demucs.separate -n htdemucs_6s -d {device} {audio_path} -o out"
         process = subprocess.run(command, shell=True, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-        print("Demucs script output:", process.stdout.decode())
     except subprocess.CalledProcessError as e:
-        print("Error in Demucs script:", e.stderr.decode())
-        return [gr.Audio(visible=False)] * 8 + [loading_gif_path]
+        return None  # Handle the error appropriately
 
     try:
         separator = Separator("./out/htdemucs_6s/test/vocals.wav", model_name='UVR_MDXNET_KARA_2', use_cuda=device=='cuda', output_format='wav')
         primary_stem_path, secondary_stem_path = separator.separate()
     except Exception as e:
-        print("Error in custom separation:", str(e))
-        return [gr.Audio(visible=False)] * 8 + [loading_gif_path]
+        return None  # Handle the error appropriately
 
+    # Generate paths for the stems
     stem_paths = {
         "vocals": "./out/htdemucs_6s/test/vocals.wav" if vocals else None,
         "bass": "./out/htdemucs_6s/test/bass.wav" if bass else None,
@@ -178,9 +184,23 @@ def inference(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, bac
         "backing_vocals": secondary_stem_path if backing_vocals else None
     }
 
-    # Once processing is done, hide the GIF by returning a transparent image
-    gr.Image(visible=False)
-    return [gr.Audio(stem_paths[stem], visible=bool(stem_paths[stem])) for stem in stem_paths]  
+    return [stem_paths[stem] for stem in stem_paths]
+
+def inference(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, backing_vocals):
+    global show_loading_gif
+
+    # Start loading
+    start_loading()
+
+    # Call the main processing function
+    audio_outputs = process_audio(audio, vocals, bass, drums, other, piano, guitar, lead_vocals, backing_vocals)
+
+    # Stop loading
+    stop_loading()
+
+    # Return the outputs along with the loading GIF state
+    loading_gif_path = "7RwF.gif" if show_loading_gif else ""
+    return loading_gif_path, audio_outputs
 
 # Define checkboxes for each stem
 checkbox_labels = ["Full Vocals", "Bass", "Drums", "Other", "Piano", "Guitar", "Lead Vocals", "Backing Vocals"]
@@ -192,9 +212,10 @@ description = "Music Source Separation in the Waveform Domain. Upload your audio
 iface = gr.Interface(
     inference, 
     [gr.components.Audio(type="numpy", label="Input")] + checkboxes,
-    [gr.Audio(label=label, visible=False) for label in checkbox_labels],
+    [gr.Image(), [gr.Audio(label=label, visible=False) for label in checkbox_labels]],
     title=title,
     description=description,
 )
 
 iface.launch()
+