fix code style

egs2/TEMPLATE/asr1/asr.sh

@@ -755,7 +755,7 @@ if ! "${skip_train}"; then
             log "LM collect-stats started... log: '${_logdir}/stats.*.log'"
             # NOTE: --*_shape_file doesn't require length information if --batch_type=unsorted,
             #       but it's used only for deciding the sample ids.
-            # shellcheck disable=SC2086
+            # shellcheck disable=SC2046,SC2086
             ${train_cmd} JOB=1:"${_nj}" "${_logdir}"/stats.JOB.log \
                 ${python} -m espnet2.bin.lm_train \
                     --collect_stats true \
@@ -967,7 +967,7 @@ if ! "${skip_train}"; then
         # NOTE: --*_shape_file doesn't require length information if --batch_type=unsorted,
         #       but it's used only for deciding the sample ids.
 
-        # shellcheck disable=SC2086
+        # shellcheck disable=SC2046,SC2086
         ${train_cmd} JOB=1:"${_nj}" "${_logdir}"/stats.JOB.log \
             ${python} -m espnet2.bin.asr_train \
                 --collect_stats true \
@@ -1242,7 +1242,7 @@ if ! "${skip_eval}"; then
 
             # 2. Submit decoding jobs
             log "Decoding started... log: '${_logdir}/asr_inference.*.log'"
-            # shellcheck disable=SC2086
+            # shellcheck disable=SC2046,SC2086
             ${_cmd} --gpu "${_ngpu}" JOB=1:"${_nj}" "${_logdir}"/asr_inference.JOB.log \
                 ${python} -m ${asr_inference_tool} \
                     --batch_size ${batch_size} \

espnet/asr/pytorch_backend/asr.py

@@ -4,12 +4,12 @@
 """Training/decoding definition for the speech recognition task."""
 
 import copy
-from packaging.version import parse as V
 import itertools
 import json
 import logging
 import math
 import os
+from packaging.version import parse as V
 
 from chainer import reporter as reporter_module
 from chainer import training
@@ -999,9 +999,7 @@ def recog(args):
 
         # Dunno why but weight_observer from dynamic quantized module must have
         # dtype=torch.qint8 with torch < 1.5 although dtype=torch.float16 is supported.
-        if args.quantize_dtype == "float16" and torch.__version__ < V(
-            "1.5.0"
-        ):
+        if args.quantize_dtype == "float16" and torch.__version__ < V("1.5.0"):
             raise ValueError(
                 "float16 dtype for dynamic quantization is not supported with torch "
                 "version < 1.5.0. Switching to qint8 dtype instead."

espnet/asr/pytorch_backend/recog.py

@@ -62,9 +62,7 @@ def recog_v2(args):
                 "Quantized LSTM in ESPnet is only supported with torch 1.4+."
             )
 
-        if args.quantize_dtype == "float16" and torch.__version__ < V(
-            "1.5.0"
-        ):
+        if args.quantize_dtype == "float16" and torch.__version__ < V("1.5.0"):
             raise ValueError(
                 "float16 dtype for dynamic quantization is not supported with torch "
                 "version < 1.5.0. Switching to qint8 dtype instead."

espnet/nets/pytorch_backend/ctc.py

@@ -1,5 +1,5 @@
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 
 import numpy as np
 import six
@@ -28,11 +28,7 @@ def __init__(self, odim, eprojs, dropout_rate, ctc_type="warpctc", reduce=True):
         self.probs = None  # for visualization
 
         # In case of Pytorch >= 1.7.0, CTC will be always builtin
-        self.ctc_type = (
-            ctc_type
-            if V(torch.__version__) < V("1.7.0")
-            else "builtin"
-        )
+        self.ctc_type = ctc_type if V(torch.__version__) < V("1.7.0") else "builtin"
 
         if ctc_type != self.ctc_type:
             logging.warning(f"CTC was set to {self.ctc_type} due to PyTorch version.")

espnet/nets/pytorch_backend/e2e_tts_fastspeech.py

@@ -576,7 +576,7 @@ def _forward(
         alpha=1.0,
     ):
         # forward encoder
-        x_masks = self._source_mask(ilens)
+        x_masks = self._source_mask(ilens).to(xs.device)
         hs, _ = self.encoder(xs, x_masks)  # (B, Tmax, adim)
 
         # integrate speaker embedding
@@ -603,7 +603,7 @@ def _forward(
                 olens_in = olens.new([olen // self.reduction_factor for olen in olens])
             else:
                 olens_in = olens
-            h_masks = self._source_mask(olens_in)
+            h_masks = self._source_mask(olens_in).to(xs.device)
         else:
             h_masks = None
         zs, _ = self.decoder(hs, h_masks)  # (B, Lmax, adim)
@@ -816,7 +816,7 @@ def _source_mask(self, ilens):
                      [1, 1, 1, 0, 0]]], dtype=torch.uint8)
 
         """
-        x_masks = make_non_pad_mask(ilens).to(next(self.parameters()).device)
+        x_masks = make_non_pad_mask(ilens)
         return x_masks.unsqueeze(-2)
 
     def _load_teacher_model(self, model_path):

espnet/nets/pytorch_backend/e2e_vc_transformer.py

@@ -673,7 +673,7 @@ def forward(self, xs, ilens, ys, labels, olens, spembs=None, *args, **kwargs):
             xs_ds, ilens_ds = xs, ilens
 
         # forward encoder
-        x_masks = self._source_mask(ilens_ds)
+        x_masks = self._source_mask(ilens_ds).to(xs.device)
         hs, hs_masks = self.encoder(xs_ds, x_masks)
 
         # integrate speaker embedding
@@ -701,7 +701,7 @@ def forward(self, xs, ilens, ys, labels, olens, spembs=None, *args, **kwargs):
             ilens_ds_st = ilens_ds
 
         # forward decoder
-        y_masks = self._target_mask(olens_in)
+        y_masks = self._target_mask(olens_in).to(xs.device)
         zs, _ = self.decoder(ys_in, y_masks, hs_int, hs_masks)
         # (B, Lmax//r, odim * r) -> (B, Lmax//r * r, odim)
         before_outs = self.feat_out(zs).view(zs.size(0), -1, self.odim)
@@ -977,7 +977,7 @@ def calculate_all_attentions(
                 xs_ds, ilens_ds = xs, ilens
 
             # forward encoder
-            x_masks = self._source_mask(ilens_ds)
+            x_masks = self._source_mask(ilens_ds).to(xs.device)
             hs, hs_masks = self.encoder(xs_ds, x_masks)
 
             # integrate speaker embedding
@@ -996,7 +996,7 @@ def calculate_all_attentions(
             ys_in = self._add_first_frame_and_remove_last_frame(ys_in)
 
             # forward decoder
-            y_masks = self._target_mask(olens_in)
+            y_masks = self._target_mask(olens_in).to(xs.device)
             zs, _ = self.decoder(ys_in, y_masks, hs, hs_masks)
 
             # calculate final outputs
@@ -1099,7 +1099,7 @@ def _source_mask(self, ilens):
                     [[1, 1, 1, 0, 0]]], dtype=torch.uint8)
 
         """
-        x_masks = make_non_pad_mask(ilens).to(next(self.parameters()).device)
+        x_masks = make_non_pad_mask(ilens)
         return x_masks.unsqueeze(-2)
 
     def _target_mask(self, olens):
@@ -1128,7 +1128,7 @@ def _target_mask(self, olens):
                      [1, 1, 1, 0, 0]]], dtype=torch.uint8)
 
         """
-        y_masks = make_non_pad_mask(olens).to(next(self.parameters()).device)
+        y_masks = make_non_pad_mask(olens)
         s_masks = subsequent_mask(y_masks.size(-1), device=y_masks.device).unsqueeze(0)
         return y_masks.unsqueeze(-2) & s_masks
 

espnet/nets/pytorch_backend/nets_utils.py

@@ -151,10 +151,7 @@ def make_pad_mask(lengths, xs=None, length_dim=-1, maxlen=None):
         raise ValueError("length_dim cannot be 0: {}".format(length_dim))
 
     if not isinstance(lengths, list):
-        lengths = lengths.tolist()
-    else:
-        assert isinstance(lengths, torch.Tensor), type(lengths)
-        lengths = lengths.long()
+        lengths = lengths.long().tolist()
 
     bs = int(len(lengths))
     if maxlen is None:

espnet2/asr/espnet_model.py

@@ -1,6 +1,6 @@
 from contextlib import contextmanager
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 from typing import Dict
 from typing import List
 from typing import Optional

espnet2/asr/maskctc_model.py

@@ -1,7 +1,7 @@
 from contextlib import contextmanager
-from packaging.version import parse as V
 from itertools import groupby
 import logging
+from packaging.version import parse as V
 from typing import Dict
 from typing import List
 from typing import Optional

espnet2/diar/espnet_model.py

@@ -2,8 +2,8 @@
 #  Apache 2.0  (http://www.apache.org/licenses/LICENSE-2.0)
 
 from contextlib import contextmanager
-from packaging.version import parse as V
 from itertools import permutations
+from packaging.version import parse as V
 from typing import Dict
 from typing import Optional
 from typing import Tuple

espnet2/enh/espnet_enh_s2t_model.py

@@ -1,6 +1,6 @@
 from contextlib import contextmanager
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 import random
 from typing import Dict
 from typing import List

espnet2/enh/loss/criterions/tf_domain.py

@@ -1,8 +1,8 @@
 from abc import ABC
 from abc import abstractmethod
-from packaging.version import parse as V
 from functools import reduce
 import math
+from packaging.version import parse as V
 
 import torch
 import torch.nn.functional as F

espnet2/mt/espnet_model.py

@@ -1,6 +1,6 @@
 from contextlib import contextmanager
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 from typing import Dict
 from typing import List
 from typing import Optional

espnet2/st/espnet_model.py

@@ -1,6 +1,6 @@
 from contextlib import contextmanager
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 from typing import Dict
 from typing import List
 from typing import Optional

espnet2/tasks/abs_task.py

@@ -3,10 +3,10 @@
 from abc import abstractmethod
 import argparse
 from dataclasses import dataclass
-from packaging.version import parse as V
 import functools
 import logging
 import os
+from packaging.version import parse as V
 from pathlib import Path
 import sys
 from typing import Any

espnet2/train/reporter.py

@@ -3,8 +3,8 @@
 from contextlib import contextmanager
 import dataclasses
 import datetime
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 from pathlib import Path
 import time
 from typing import ContextManager

espnet2/train/trainer.py

@@ -3,8 +3,8 @@
 from contextlib import contextmanager
 import dataclasses
 from dataclasses import is_dataclass
-from packaging.version import parse as V
 import logging
+from packaging.version import parse as V
 from pathlib import Path
 import time
 from typing import Dict

espnet2/utils/griffin_lim.py

@@ -7,8 +7,8 @@
 
 import logging
 
-from packaging.version import parse as V
 from functools import partial
+from packaging.version import parse as V
 from typeguard import check_argument_types
 from typing import Optional