Delete unused variables

.gitignore

@@ -156,6 +156,7 @@ filelists/test.txt
 filelists/train.txt
 filelists/val.txt
 .idea/
+.vscode/
 .idea/modules.xml
 .idea/so-vits-svc.iml
 .idea/vcs.xml
@@ -168,3 +169,4 @@ pretrain/vec-256-layer-9.onnx
 pretrain/vec-256-layer-12.onnx
 pretrain/vec-768-layer-9.onnx
 .vscode/launch.json
+.ruff.toml

.ruff.toml

@@ -1,4 +0,0 @@
-select = ["E", "F", "I"]
-
-# Never enforce `E501` (line length violations).
-ignore = ["E501"]

.vscode/extensions.json

@@ -1,6 +0,0 @@
-{
-    "recommendations": [
-        "charliermarsh.ruff",
-        "ms-python.python"
-    ]
-}

diffusion/data_loaders.py

@@ -131,7 +131,6 @@ class AudioDataset(Dataset):
         with open(filelists,"r") as f:
             self.paths = f.read().splitlines()
         for name_ext in tqdm(self.paths, total=len(self.paths)):
-            os.path.splitext(name_ext)[0]
             path_audio = name_ext
             duration = librosa.get_duration(filename = path_audio, sr = self.sample_rate)
             

diffusion/diffusion_onnx.py

@@ -581,9 +581,6 @@ class GaussianDiffusion(nn.Module):
         plms_noise_stage = torch.tensor(0, dtype=torch.long, device=device)
         noise_list = torch.zeros((0, 1, 1, self.mel_bins, n_frames), device=device)
 
-        ot = step_range[0]
-        torch.full((1,), ot, device=device, dtype=torch.long)
-
         for t in step_range:
             t_1 = torch.full((1,), t, device=device, dtype=torch.long)
             noise_pred = self.denoise_fn(x, t_1, cond)

diffusion/dpm_solver_pytorch.py

@@ -557,7 +557,6 @@ class DPM_Solver:
             x_t: A pytorch tensor. The approximated solution at time `t`.
         """
         ns = self.noise_schedule
-        x.dim()
         lambda_s, lambda_t = ns.marginal_lambda(s), ns.marginal_lambda(t)
         h = lambda_t - lambda_s
         log_alpha_s, log_alpha_t = ns.marginal_log_mean_coeff(s), ns.marginal_log_mean_coeff(t)

diffusion/unit2mel.py

@@ -121,7 +121,6 @@ class Unit2Mel(nn.Module):
         hubert_hidden_size = self.input_channel
         n_frames = 10
         hubert = torch.randn((1, n_frames, hubert_hidden_size))
-        torch.arange(end=n_frames).unsqueeze(0).long()
         f0 = torch.randn((1, n_frames))
         volume = torch.randn((1, n_frames))
         spks = {}

onnx_export_speaker_mix.py

@@ -130,7 +130,6 @@ def main():
         "Characters": spklist
     }
 
-    json.dumps(MoeVSConf)
     with open(f"checkpoints/{path}.json", 'w') as MoeVsConfFile:
         json.dump(MoeVSConf, MoeVsConfFile, indent = 4)
 

vdecoder/hifigan/models.py

@@ -201,8 +201,6 @@ class SineGen(torch.nn.Module):
         output uv: tensor(batchsize=1, length, 1)
         """
         with torch.no_grad():
-            torch.zeros(f0.shape[0], f0.shape[1], self.dim,
-                                 device=f0.device)
             # fundamental component
             fn = torch.multiply(f0, torch.FloatTensor([[range(1, self.harmonic_num + 2)]]).to(f0.device))
 

vdecoder/hifiganwithsnake/models.py

@@ -214,8 +214,6 @@ class SineGen(torch.nn.Module):
         output uv: tensor(batchsize=1, length, 1)
         """
         with torch.no_grad():
-            torch.zeros(f0.shape[0], f0.shape[1], self.dim,
-                                 device=f0.device)
             # fundamental component
             fn = torch.multiply(f0, torch.FloatTensor([[range(1, self.harmonic_num + 2)]]).to(f0.device))