Merge branch 'clean' of https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI into clean

infer/modules/uvr5/mdxnet.py

@@ -13,7 +13,7 @@ cpu = torch.device("cpu")
 
 class ConvTDFNetTrim:
     def __init__(
-            self, device, model_name, target_name, L, dim_f, dim_t, n_fft, hop=1024
+        self, device, model_name, target_name, L, dim_f, dim_t, n_fft, hop=1024
     ):
         super(ConvTDFNetTrim, self).__init__()
 
@@ -83,7 +83,7 @@ def get_models(device, dim_f, dim_t, n_fft):
         dim_f=dim_f,
         dim_t=dim_t,
         n_fft=n_fft,
-        )
+    )
 
 
 class Predictor:
@@ -95,7 +95,11 @@ class Predictor:
         )
         self.model = ort.InferenceSession(
             os.path.join(args.onnx, self.model_.target_name + ".onnx"),
-            providers=["CUDAExecutionProvider", "DmlExecutionProvider", "CPUExecutionProvider"],
+            providers=[
+                "CUDAExecutionProvider",
+                "DmlExecutionProvider",
+                "CPUExecutionProvider",
+            ],
         )
         print("onnx load done")
 
@@ -236,4 +240,4 @@ class MDXNetDereverb:
         self.device = device
 
     def path_audio(self, input, vocal_root, others_root, format):
-        self.pred.prediction(input, vocal_root, others_root, format)
+        self.pred.prediction(input, vocal_root, others_root, format)

infer/modules/uvr5/modules.py

@@ -27,7 +27,9 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
             func = AudioPre if "DeEcho" not in model_name else AudioPreDeEcho
             pre_fun = func(
                 agg=int(agg),
-                model_path=os.path.join(os.getenv("weight_uvr5_root"), model_name + ".pth"),
+                model_path=os.path.join(
+                    os.getenv("weight_uvr5_root"), model_name + ".pth"
+                ),
                 device=config.device,
                 is_half=config.is_half,
             )
@@ -54,7 +56,10 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
                 need_reformat = 1
                 traceback.print_exc()
             if need_reformat == 1:
-                tmp_path = "%s/%s.reformatted.wav" % (os.path.join("tmp"), os.path.basename(inp_path))
+                tmp_path = "%s/%s.reformatted.wav" % (
+                    os.path.join("tmp"),
+                    os.path.basename(inp_path),
+                )
                 os.system(
                     "ffmpeg -i %s -vn -acodec pcm_s16le -ac 2 -ar 44100 %s -y"
                     % (inp_path, tmp_path)
@@ -89,4 +94,3 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
         if torch.cuda.is_available():
             torch.cuda.empty_cache()
     yield "\n".join(infos)
-    

infer/modules/uvr5/preprocess.py

@@ -205,7 +205,7 @@ class AudioPreDeEcho:
         self.model = model
 
     def _path_audio_(
-            self, music_file, vocal_root=None, ins_root=None, format="flac"
+        self, music_file, vocal_root=None, ins_root=None, format="flac"
     ):  # 3个VR模型vocal和ins是反的
         if ins_root is None and vocal_root is None:
             return "No save root."
@@ -222,7 +222,7 @@ class AudioPreDeEcho:
             if d == bands_n:  # high-end band
                 (
                     X_wave[d],
-                            _,
+                    _,
                 ) = librosa.core.load(  # 理论上librosa读取可能对某些音频有bug，应该上ffmpeg读取，但是太麻烦了弃坑
                     music_file,
                     bp["sr"],
@@ -341,4 +341,4 @@ class AudioPreDeEcho:
                     os.system(
                         "ffmpeg -i %s -vn %s -q:a 2 -y"
                         % (path, path[:-4] + ".%s" % format)
-                    )
+                    )