Merge branch 'tmp_muskit' of github.com:A-Quarter-Mile/espnet into tm…

…p_muskit

egs2/TEMPLATE/svs1/svs.sh

@@ -114,7 +114,7 @@ lang=noinfo      # The language type of corpus.
 text_fold_length=150   # fold_length for text data.
 singing_fold_length=800 # fold_length for singing data.
 
-# Upload model related 
+# Upload model related
 hf_repo=
 
 help_message=$(cat << EOF

espnet2/bin/svs_inference.py

@@ -92,7 +92,8 @@ def __init__(
         decode_conf.update({"use_teacher_forcing": use_teacher_forcing})
         if isinstance(self.svs, VITS):
             decode_conf.update(
-                noise_scale=noise_scale, noise_scale_dur=noise_scale_dur,
+                noise_scale=noise_scale,
+                noise_scale_dur=noise_scale_dur,
             )
         self.decode_conf = decode_conf
 
@@ -364,7 +365,8 @@ def inference(
     from matplotlib.ticker import MaxNLocator
 
     with NpyScpWriter(
-        output_dir / "norm", output_dir / "norm/feats.scp",
+        output_dir / "norm",
+        output_dir / "norm/feats.scp",
     ) as norm_writer, NpyScpWriter(
         output_dir / "denorm", output_dir / "denorm/feats.scp"
     ) as denorm_writer, open(
@@ -524,13 +526,22 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--output_dir", type=str, required=True, help="The path of output directory",
+        "--output_dir",
+        type=str,
+        required=True,
+        help="The path of output directory",
     )
     parser.add_argument(
-        "--ngpu", type=int, default=0, help="The number of gpus. 0 indicates CPU mode",
+        "--ngpu",
+        type=int,
+        default=0,
+        help="The number of gpus. 0 indicates CPU mode",
     )
     parser.add_argument(
-        "--seed", type=int, default=0, help="Random seed",
+        "--seed",
+        type=int,
+        default=0,
+        help="Random seed",
     )
     parser.add_argument(
         "--dtype",
@@ -545,7 +556,10 @@ def get_parser():
         help="The number of workers used for DataLoader",
     )
     parser.add_argument(
-        "--batch_size", type=int, default=1, help="The batch size for inference",
+        "--batch_size",
+        type=int,
+        default=1,
+        help="The batch size for inference",
     )
 
     group = parser.add_argument_group("Input data related")
@@ -556,18 +570,25 @@ def get_parser():
         action="append",
     )
     group.add_argument(
-        "--key_file", type=str_or_none,
+        "--key_file",
+        type=str_or_none,
     )
     group.add_argument(
-        "--allow_variable_data_keys", type=str2bool, default=False,
+        "--allow_variable_data_keys",
+        type=str2bool,
+        default=False,
     )
 
     group = parser.add_argument_group("The model configuration related")
     group.add_argument(
-        "--train_config", type=str, help="Training configuration file.",
+        "--train_config",
+        type=str,
+        help="Training configuration file.",
     )
     group.add_argument(
-        "--model_file", type=str, help="Model parameter file.",
+        "--model_file",
+        type=str,
+        help="Model parameter file.",
     )
 
     group = parser.add_argument_group("Decoding related")

espnet2/train/preprocessor.py

@@ -1053,7 +1053,8 @@ def __init__(
                 g2p_type=g2p_type,
             )
             self.token_id_converter = TokenIDConverter(
-                token_list=token_list, unk_symbol=unk_symbol,
+                token_list=token_list,
+                unk_symbol=unk_symbol,
             )
         else:
             self.text_cleaner = None
@@ -1073,10 +1074,7 @@ def __call__(
                 ma = np.max(np.abs(singing))
                 data[self.singing_name] = singing * self.singing_volume_normalize / ma
 
-        if (
-            self.midi_name in data
-            and self.label_name in data
-        ):
+        if self.midi_name in data and self.label_name in data:
             # Load label info
             lab_timeseq, text = data[self.label_name]
             lab_len = len(text)