You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
91 lines
3.0 KiB
91 lines
3.0 KiB
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
"""
|
|
Load nnet3 training egs which generated by kaldi
|
|
"""
|
|
|
|
import random
|
|
import numpy as np
|
|
import kaldi_python_io as k_io
|
|
from paddle.io import Dataset
|
|
import paddlespeech.vector.utils.utils as utils
|
|
from paddlespeech.vector import _logger as log
|
|
class KaldiEgsDataset(Dataset):
|
|
"""
|
|
Dataset used to load kaldi nnet3 egs files.
|
|
"""
|
|
def __init__(self, egs_list_file, egs_idx, transforms=None):
|
|
self.scp_reader = None
|
|
self.subset_idx = egs_idx - 1
|
|
self.transforms = transforms
|
|
if not utils.is_exist(egs_list_file):
|
|
return
|
|
|
|
self.egs_files = []
|
|
with open(egs_list_file, 'r') as in_fh:
|
|
for line in in_fh:
|
|
if line.strip():
|
|
self.egs_files.append(line.strip())
|
|
|
|
self.next_subset()
|
|
|
|
def next_subset(self, target_index=None, delta_index=None):
|
|
"""
|
|
Use next specific subset
|
|
|
|
Args:
|
|
target_index: target egs index
|
|
delta_index: incremental value of egs index
|
|
"""
|
|
if self.egs_files:
|
|
if target_index:
|
|
self.subset_idx = target_index
|
|
else:
|
|
delta_index = delta_index if delta_index else 1
|
|
self.subset_idx += delta_index
|
|
log.info("egs dataset subset index: %d" % (self.subset_idx))
|
|
egs_file = self.egs_files[self.subset_idx % len(self.egs_files)]
|
|
if utils.is_exist(egs_file):
|
|
self.scp_reader = k_io.Nnet3EgsScriptReader(egs_file)
|
|
else:
|
|
log.warning("No such file or directory: %s" % (egs_file))
|
|
|
|
def __getitem__(self, index):
|
|
if self.scp_reader is None:
|
|
return {}
|
|
index %= len(self)
|
|
in_dict, out_dict = self.scp_reader[index]
|
|
x = np.array(in_dict['matrix'])
|
|
x = np.transpose(x)
|
|
y = np.array(out_dict['matrix'][0][0][0], dtype=np.int).reshape((1,))
|
|
if self.transforms is not None:
|
|
idx = random.randint(0, len(self.transforms) - 1)
|
|
x = self.transforms[idx](x)
|
|
return x, y
|
|
|
|
def __len__(self):
|
|
return len(self.scp_reader)
|
|
|
|
def __iter__(self):
|
|
self._start = 0
|
|
return self
|
|
|
|
def __next__(self):
|
|
if self._start < len(self):
|
|
ret = self[self._start]
|
|
self._start += 1
|
|
return ret
|
|
else:
|
|
raise StopIteration |