Update app.py

app.py

@@ -4,7 +4,7 @@ import shutil
 import logging
 import gradio as gr
 
-from audio_separator.separator import Separator
+from PolUVR.separator import Separator
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 use_autocast = device == "cuda"
@@ -18,11 +18,13 @@ ROFORMER_MODELS = {
     'BS-Roformer-Viperx-1297': 'model_bs_roformer_ep_317_sdr_12.9755.ckpt',
     'BS-Roformer-De-Reverb': 'deverb_bs_roformer_8_384dim_10depth.ckpt',
     'Mel-Roformer-Viperx-1143': 'model_mel_band_roformer_ep_3005_sdr_11.4360.ckpt',
+    'Mel-Roformer-Vocals-Kim': 'vocals_mel_band_roformer.ckpt',
     'Mel-Roformer-Crowd-Aufr33-Viperx': 'mel_band_roformer_crowd_aufr33_viperx_sdr_8.7144.ckpt',
     'Mel-Roformer-Karaoke-Aufr33-Viperx': 'mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt',
     'Mel-Roformer-Denoise-Aufr33': 'denoise_mel_band_roformer_aufr33_sdr_27.9959.ckpt',
     'Mel-Roformer-Denoise-Aufr33-Aggr': 'denoise_mel_band_roformer_aufr33_aggr_sdr_27.9768.ckpt',
     'MelBand Roformer Kim | Inst V1 by Unwa': 'melband_roformer_inst_v1.ckpt',
+    'MelBand Roformer Kim | Inst V1 (E) by Unwa': 'melband_roformer_inst_v1e.ckpt',
     'MelBand Roformer Kim | Inst V2 by Unwa': 'melband_roformer_inst_v2.ckpt',
     'MelBand Roformer Kim | InstVoc Duality V1 by Unwa': 'melband_roformer_instvoc_duality_v1.ckpt',
     'MelBand Roformer Kim | InstVoc Duality V2 by Unwa': 'melband_roformer_instvox_duality_v2.ckpt',
@@ -124,7 +126,7 @@ def print_message(input_file, model_name):
     """Prints information about the audio separation process."""
     base_name = os.path.splitext(os.path.basename(input_file))[0]
     print("\n")
-    print("🎵 Audio Separator 🎵")
+    print("🎵 PolUVR 🎵")
     print("Input audio:", base_name)
     print("Separation Model:", model_name)
     print("Audio Separation Process...")
@@ -344,14 +346,14 @@ def update_stems(model):
         return gr.update(visible=False)
 
 with gr.Blocks(
-    title="🎵 Audio Separator 🎵",
+    title="🎵 PolUVR 🎵",
     css="footer{display:none !important}",
     theme=gr.themes.Default(
         spacing_size="sm",
         radius_size="lg",
     )
 ) as app:
-    gr.HTML("<h1> 🎵 Audio Separator 🎵 </h1>")
+    gr.HTML("<h1> 🎵 PolUVR 🎵 </h1>")
 
     with gr.Tab("Roformer"):
         with gr.Group():
@@ -361,7 +363,7 @@ with gr.Blocks(
                 roformer_seg_size = gr.Slider(minimum=32, maximum=4000, step=32, value=256, label="Segment Size", info="Larger consumes more resources, but may give better results.")
                 roformer_override_seg_size = gr.Checkbox(value=False, label="Override segment size", info="Override model default segment size instead of using the model default value.")
                 roformer_overlap = gr.Slider(minimum=2, maximum=10, step=1, value=8, label="Overlap", info="Amount of overlap between prediction windows. Lower is better but slower.")
-                roformer_pitch_shift = gr.Slider(minimum=-12, maximum=12, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.")
+                roformer_pitch_shift = gr.Slider(minimum=-24, maximum=24, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.")
         with gr.Row():
             roformer_audio = gr.Audio(label="Input Audio", type="filepath")
         with gr.Row():
@@ -378,7 +380,7 @@ with gr.Blocks(
                 mdx23c_seg_size = gr.Slider(minimum=32, maximum=4000, step=32, value=256, label="Segment Size", info="Larger consumes more resources, but may give better results.")
                 mdx23c_override_seg_size = gr.Checkbox(value=False, label="Override segment size", info="Override model default segment size instead of using the model default value.")
                 mdx23c_overlap = gr.Slider(minimum=2, maximum=50, step=1, value=8, label="Overlap", info="Amount of overlap between prediction windows. Higher is better but slower.")
-                mdx23c_pitch_shift = gr.Slider(minimum=-12, maximum=12, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.")
+                mdx23c_pitch_shift = gr.Slider(minimum=-24, maximum=24, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.")
         with gr.Row():
             mdx23c_audio = gr.Audio(label="Input Audio", type="filepath")
         with gr.Row():
@@ -410,7 +412,7 @@ with gr.Blocks(
                 vr_model = gr.Dropdown(label="Select the Model", choices=VR_ARCH_MODELS)
             with gr.Row():
                 vr_window_size = gr.Slider(minimum=320, maximum=1024, step=32, value=512, label="Window Size", info="Balance quality and speed. 1024 = fast but lower, 320 = slower but better quality.")
-                vr_aggression = gr.Slider(minimum=1, maximum=50, step=1, value=5, label="Agression", info="Intensity of primary stem extraction.")
+                vr_aggression = gr.Slider(minimum=1, maximum=100, step=1, value=5, label="Agression", info="Intensity of primary stem extraction.")
                 vr_tta = gr.Checkbox(value=False, label="TTA", info="Enable Test-Time-Augmentation; slow but improves quality.")
                 vr_post_process = gr.Checkbox(value=False, label="Post Process", info="Identify leftover artifacts within vocal output; may improve separation for some songs.")
                 vr_post_process_threshold = gr.Slider(minimum=0.1, maximum=0.3, step=0.1, value=0.2, label="Post Process Threshold", info="Threshold for post-processing.")
@@ -468,7 +470,6 @@ with gr.Blocks(
                 * **NAME** - Input File Name
                 * **STEM** - Stem Name (e.g., Vocals, Instrumental)
                 * **MODEL** - Model Name (e.g., BS-Roformer-Viperx-1297)
-
                 Example:
                 * Usage: NAME_(STEM)_MODEL
                 * Output File Name: Music_(Vocals)_BS-Roformer-Viperx-1297