Update app.py

app.py

@@ -1,4 +1,5 @@
 import os
+import shutil
 import logging
 import gradio as gr
 
@@ -108,9 +109,19 @@ def rename_stems(input_file, output_dir, stems, output_format):
         renamed_stems.append(new_path)
     return renamed_stems
 
+def prepare_output_dir(input_file, output_dir):
+    """Create a directory for the output files and clean it if it already exists."""
+    base_name = os.path.splitext(os.path.basename(input_file))[0]
+    out_dir = os.path.join(output_dir, base_name)
+    if os.path.exists(out_dir):
+        shutil.rmtree(out_dir)
+    os.makedirs(out_dir)
+    return out_dir
+
 def roformer_separator(audio, model_key, seg_size, overlap, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()):
     """Separate audio using Roformer model."""
     model = ROFORMER_MODELS[model_key]
+    out_dir = prepare_output_dir(audio, out_dir)
     separator = Separator(
         log_level=logging.WARNING,
         model_file_dir=model_dir,
@@ -138,6 +149,7 @@ def roformer_separator(audio, model_key, seg_size, overlap, model_dir, out_dir,
 
 def mdx23c_separator(audio, model, seg_size, overlap, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()):
     """Separate audio using MDX23C model."""
+    out_dir = prepare_output_dir(audio, out_dir)
     separator = Separator(
         log_level=logging.WARNING,
         model_file_dir=model_dir,
@@ -165,6 +177,7 @@ def mdx23c_separator(audio, model, seg_size, overlap, model_dir, out_dir, out_fo
 
 def mdx_separator(audio, model, hop_length, seg_size, overlap, denoise, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()):
     """Separate audio using MDX-NET model."""
+    out_dir = prepare_output_dir(audio, out_dir)
     separator = Separator(
         log_level=logging.WARNING,
         model_file_dir=model_dir,
@@ -194,6 +207,7 @@ def mdx_separator(audio, model, hop_length, seg_size, overlap, denoise, model_di
 
 def vr_separator(audio, model, window_size, aggression, tta, post_process, post_process_threshold, high_end_process, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()):
     """Separate audio using VR ARCH model."""
+    out_dir = prepare_output_dir(audio, out_dir)
     separator = Separator(
         log_level=logging.WARNING,
         model_file_dir=model_dir,
@@ -225,6 +239,7 @@ def vr_separator(audio, model, window_size, aggression, tta, post_process, post_
 
 def demucs_separator(audio, model, seg_size, shifts, overlap, segments_enabled, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()):
     """Separate audio using Demucs model."""
+    out_dir = prepare_output_dir(audio, out_dir)
     separator = Separator(
         log_level=logging.WARNING,
         model_file_dir=model_dir,
@@ -267,7 +282,7 @@ with gr.Blocks(
         with gr.Group():
             model_file_dir = gr.Textbox(value="/tmp/audio-separator-models/", label="Directory for storing model files", placeholder="/tmp/audio-separator-models/", interactive=False)
             with gr.Row():
-                output_dir = gr.Textbox(value="output", label="File output directory", placeholder="output", interactive=True)
+                output_dir = gr.Textbox(value="output", label="File output directory", placeholder="output", interactive=False)
                 output_format = gr.Dropdown(value="wav", choices=["wav", "flac", "mp3"], label="Output Format")
             with gr.Row():
                 norm_threshold = gr.Slider(value=0.9, step=0.1, minimum=0, maximum=1, label="Normalization", info="max peak amplitude to normalize input and output audio.")