refactor io, loss, conv, rnn, gradclip, model, utils

pull/538/head
Hui Zhang 5 years ago
parent 1ae41eac90
commit 4da2852982

@ -0,0 +1,9 @@
ThreadPool/
build/
dist/
kenlm/
openfst-1.6.3/
openfst-1.6.3.tar.gz
swig_decoders.egg-info/
decoders_wrap.cxx
swig_decoders.py

@ -13,6 +13,7 @@
# limitations under the License.
from yacs.config import CfgNode as CN
from deepspeech.models.DeepSpeech2 import DeepSpeech2Model
_C = CN()
_C.data = CN(
@ -50,6 +51,8 @@ _C.model = CN(
share_rnn_weights=True #Whether to share input-hidden weights between forward and backward directional RNNs.Notice that for GRU, weight sharing is not supported.
))
DeepSpeech2Model.params(_C.model)
_C.training = CN(
dict(
lr=5e-4, # learning rate

@ -1,584 +0,0 @@
# 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.
import math
import random
import tarfile
import logging
import numpy as np
from collections import namedtuple
from functools import partial
import paddle
from paddle.io import Dataset
from paddle.io import DataLoader
from paddle.io import BatchSampler
from paddle.io import DistributedBatchSampler
from paddle import distributed as dist
from deepspeech.frontend.utility import read_manifest
from deepspeech.frontend.augmentor.augmentation import AugmentationPipeline
from deepspeech.frontend.featurizer.speech_featurizer import SpeechFeaturizer
from deepspeech.frontend.speech import SpeechSegment
from deepspeech.frontend.normalizer import FeatureNormalizer
logger = logging.getLogger(__name__)
__all__ = [
"DeepSpeech2Dataset",
"DeepSpeech2DistributedBatchSampler",
"DeepSpeech2BatchSampler",
"SpeechCollator",
]
class DeepSpeech2Dataset(Dataset):
def __init__(self,
manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config='{}',
max_duration=float('inf'),
min_duration=0.0,
stride_ms=10.0,
window_ms=20.0,
n_fft=None,
max_freq=None,
target_sample_rate=16000,
specgram_type='linear',
use_dB_normalization=True,
target_dB=-20,
random_seed=0,
keep_transcription_text=False):
super().__init__()
self._max_duration = max_duration
self._min_duration = min_duration
self._normalizer = FeatureNormalizer(mean_std_filepath)
self._augmentation_pipeline = AugmentationPipeline(
augmentation_config=augmentation_config, random_seed=random_seed)
self._speech_featurizer = SpeechFeaturizer(
vocab_filepath=vocab_filepath,
specgram_type=specgram_type,
stride_ms=stride_ms,
window_ms=window_ms,
n_fft=n_fft,
max_freq=max_freq,
target_sample_rate=target_sample_rate,
use_dB_normalization=use_dB_normalization,
target_dB=target_dB)
self._rng = random.Random(random_seed)
self._keep_transcription_text = keep_transcription_text
# for caching tar files info
self._local_data = namedtuple('local_data', ['tar2info', 'tar2object'])
self._local_data.tar2info = {}
self._local_data.tar2object = {}
# read manifest
self._manifest = read_manifest(
manifest_path=manifest_path,
max_duration=self._max_duration,
min_duration=self._min_duration)
self._manifest.sort(key=lambda x: x["duration"])
@property
def manifest(self):
return self._manifest
@property
def vocab_size(self):
"""Return the vocabulary size.
:return: Vocabulary size.
:rtype: int
"""
return self._speech_featurizer.vocab_size
@property
def vocab_list(self):
"""Return the vocabulary in list.
:return: Vocabulary in list.
:rtype: list
"""
return self._speech_featurizer.vocab_list
@property
def feature_size(self):
return self._speech_featurizer.feature_size
def _parse_tar(self, file):
"""Parse a tar file to get a tarfile object
and a map containing tarinfoes
"""
result = {}
f = tarfile.open(file)
for tarinfo in f.getmembers():
result[tarinfo.name] = tarinfo
return f, result
def _subfile_from_tar(self, file):
"""Get subfile object from tar.
It will return a subfile object from tar file
and cached tar file info for next reading request.
"""
tarpath, filename = file.split(':', 1)[1].split('#', 1)
if 'tar2info' not in self._local_data.__dict__:
self._local_data.tar2info = {}
if 'tar2object' not in self._local_data.__dict__:
self._local_data.tar2object = {}
if tarpath not in self._local_data.tar2info:
object, infoes = self._parse_tar(tarpath)
self._local_data.tar2info[tarpath] = infoes
self._local_data.tar2object[tarpath] = object
return self._local_data.tar2object[tarpath].extractfile(
self._local_data.tar2info[tarpath][filename])
def process_utterance(self, audio_file, transcript):
"""Load, augment, featurize and normalize for speech data.
:param audio_file: Filepath or file object of audio file.
:type audio_file: str | file
:param transcript: Transcription text.
:type transcript: str
:return: Tuple of audio feature tensor and data of transcription part,
where transcription part could be token ids or text.
:rtype: tuple of (2darray, list)
"""
if isinstance(audio_file, str) and audio_file.startswith('tar:'):
speech_segment = SpeechSegment.from_file(
self._subfile_from_tar(audio_file), transcript)
else:
speech_segment = SpeechSegment.from_file(audio_file, transcript)
self._augmentation_pipeline.transform_audio(speech_segment)
specgram, transcript_part = self._speech_featurizer.featurize(
speech_segment, self._keep_transcription_text)
specgram = self._normalizer.apply(specgram)
return specgram, transcript_part
def _instance_reader_creator(self, manifest):
"""
Instance reader creator. Create a callable function to produce
instances of data.
Instance: a tuple of ndarray of audio spectrogram and a list of
token indices for transcript.
"""
def reader():
for instance in manifest:
inst = self.process_utterance(instance["audio_filepath"],
instance["text"])
yield inst
return reader
def __len__(self):
return len(self._manifest)
def __getitem__(self, idx):
instance = self._manifest[idx]
return self.process_utterance(instance["audio_filepath"],
instance["text"])
class DeepSpeech2DistributedBatchSampler(DistributedBatchSampler):
def __init__(self,
dataset,
batch_size,
num_replicas=None,
rank=None,
shuffle=False,
drop_last=False,
sortagrad=False,
shuffle_method="batch_shuffle"):
super().__init__(dataset, batch_size, num_replicas, rank, shuffle,
drop_last)
self._sortagrad = sortagrad
self._shuffle_method = shuffle_method
def _batch_shuffle(self, indices, batch_size, clipped=False):
"""Put similarly-sized instances into minibatches for better efficiency
and make a batch-wise shuffle.
1. Sort the audio clips by duration.
2. Generate a random number `k`, k in [0, batch_size).
3. Randomly shift `k` instances in order to create different batches
for different epochs. Create minibatches.
4. Shuffle the minibatches.
:param indices: indexes. List of int.
:type indices: list
:param batch_size: Batch size. This size is also used for generate
a random number for batch shuffle.
:type batch_size: int
:param clipped: Whether to clip the heading (small shift) and trailing
(incomplete batch) instances.
:type clipped: bool
:return: Batch shuffled mainifest.
:rtype: list
"""
rng = np.random.RandomState(self.epoch)
shift_len = rng.randint(0, batch_size - 1)
batch_indices = list(zip(* [iter(indices[shift_len:])] * batch_size))
rng.shuffle(batch_indices)
batch_indices = [item for batch in batch_indices for item in batch]
assert (clipped == False)
if not clipped:
res_len = len(indices) - shift_len - len(batch_indices)
# when res_len is 0, will return whole list, len(List[-0:]) = len(List[:])
if res_len != 0:
batch_indices.extend(indices[-res_len:])
batch_indices.extend(indices[0:shift_len])
assert len(indices) == len(
batch_indices
), f"_batch_shuffle: {len(indices)} : {len(batch_indices)} : {res_len} - {shift_len}"
return batch_indices
def __iter__(self):
num_samples = len(self.dataset)
indices = np.arange(num_samples).tolist()
indices += indices[:(self.total_size - len(indices))]
assert len(indices) == self.total_size
# sort (by duration) or batch-wise shuffle the manifest
if self.shuffle:
if self.epoch == 0 and self._sortagrad:
logger.info(
f'rank: {dist.get_rank()} dataset sortagrad! epoch {self.epoch}'
)
else:
logger.info(
f'rank: {dist.get_rank()} dataset shuffle! epoch {self.epoch}'
)
if self._shuffle_method == "batch_shuffle":
indices = self._batch_shuffle(
indices, self.batch_size, clipped=False)
elif self._shuffle_method == "instance_shuffle":
np.random.RandomState(self.epoch).shuffle(indices)
else:
raise ValueError("Unknown shuffle method %s." %
self._shuffle_method)
assert len(
indices
) == self.total_size, f"batch shuffle examples error: {len(indices)} : {self.total_size}"
# subsample
def _get_indices_by_batch_size(indices):
subsampled_indices = []
last_batch_size = self.total_size % (self.batch_size * self.nranks)
assert last_batch_size % self.nranks == 0
last_local_batch_size = last_batch_size // self.nranks
for i in range(self.local_rank * self.batch_size,
len(indices) - last_batch_size,
self.batch_size * self.nranks):
subsampled_indices.extend(indices[i:i + self.batch_size])
indices = indices[len(indices) - last_batch_size:]
subsampled_indices.extend(
indices[self.local_rank * last_local_batch_size:(
self.local_rank + 1) * last_local_batch_size])
return subsampled_indices
if self.nranks > 1:
indices = _get_indices_by_batch_size(indices)
assert len(indices) == self.num_samples
_sample_iter = iter(indices)
batch_indices = []
for idx in _sample_iter:
batch_indices.append(idx)
if len(batch_indices) == self.batch_size:
logger.info(
f"rank: {dist.get_rank()} batch index: {batch_indices} ")
yield batch_indices
batch_indices = []
if not self.drop_last and len(batch_indices) > 0:
yield batch_indices
def __len__(self):
num_samples = self.num_samples
num_samples += int(not self.drop_last) * (self.batch_size - 1)
return num_samples // self.batch_size
class DeepSpeech2BatchSampler(BatchSampler):
def __init__(self,
dataset,
batch_size,
shuffle=False,
drop_last=False,
sortagrad=False,
shuffle_method="batch_shuffle"):
self.dataset = dataset
assert isinstance(batch_size, int) and batch_size > 0, \
"batch_size should be a positive integer"
self.batch_size = batch_size
assert isinstance(shuffle, bool), \
"shuffle should be a boolean value"
self.shuffle = shuffle
assert isinstance(drop_last, bool), \
"drop_last should be a boolean number"
self.drop_last = drop_last
self.epoch = 0
self.num_samples = int(math.ceil(len(self.dataset) * 1.0))
self.total_size = self.num_samples
self._sortagrad = sortagrad
self._shuffle_method = shuffle_method
def _batch_shuffle(self, indices, batch_size, clipped=False):
"""Put similarly-sized instances into minibatches for better efficiency
and make a batch-wise shuffle.
1. Sort the audio clips by duration.
2. Generate a random number `k`, k in [0, batch_size).
3. Randomly shift `k` instances in order to create different batches
for different epochs. Create minibatches.
4. Shuffle the minibatches.
:param indices: indexes. List of int.
:type indices: list
:param batch_size: Batch size. This size is also used for generate
a random number for batch shuffle.
:type batch_size: int
:param clipped: Whether to clip the heading (small shift) and trailing
(incomplete batch) instances.
:type clipped: bool
:return: Batch shuffled mainifest.
:rtype: list
"""
rng = np.random.RandomState(self.epoch)
# must shift at leat by one
shift_len = rng.randint(0, batch_size - 1)
batch_indices = list(zip(* [iter(indices[shift_len:])] * batch_size))
rng.shuffle(batch_indices)
batch_indices = [item for batch in batch_indices for item in batch]
assert (clipped == False)
if not clipped:
res_len = len(indices) - shift_len - len(batch_indices)
# when res_len is 0, will return whole list, len(List[-0:]) = len(List[:])
if res_len != 0:
batch_indices.extend(indices[-res_len:])
batch_indices.extend(indices[0:shift_len])
assert len(indices) == len(
batch_indices
), f"_batch_shuffle: {len(indices)} : {len(batch_indices)} : {res_len} - {shift_len}"
return batch_indices
def __iter__(self):
num_samples = len(self.dataset)
indices = np.arange(num_samples).tolist()
indices += indices[:(self.total_size - len(indices))]
assert len(indices) == self.total_size
# sort (by duration) or batch-wise shuffle the manifest
if self.shuffle:
if self.epoch == 0 and self._sortagrad:
logger.info(f'dataset sortagrad! epoch {self.epoch}')
else:
logger.info(f'dataset shuffle! epoch {self.epoch}')
if self._shuffle_method == "batch_shuffle":
indices = self._batch_shuffle(
indices, self.batch_size, clipped=False)
elif self._shuffle_method == "instance_shuffle":
np.random.RandomState(self.epoch).shuffle(indices)
else:
raise ValueError("Unknown shuffle method %s." %
self._shuffle_method)
assert len(
indices
) == self.total_size, f"batch shuffle examples error: {len(indices)} : {self.total_size}"
assert len(indices) == self.num_samples
_sample_iter = iter(indices)
batch_indices = []
for idx in _sample_iter:
batch_indices.append(idx)
if len(batch_indices) == self.batch_size:
logger.info(
f"rank: {dist.get_rank()} batch index: {batch_indices} ")
yield batch_indices
batch_indices = []
if not self.drop_last and len(batch_indices) > 0:
yield batch_indices
self.epoch += 1
def __len__(self):
num_samples = self.num_samples
num_samples += int(not self.drop_last) * (self.batch_size - 1)
return num_samples // self.batch_size
class SpeechCollator():
def __init__(self, padding_to=-1, is_training=True):
"""
Padding audio features with zeros to make them have the same shape (or
a user-defined shape) within one bach.
If ``padding_to`` is -1, the maximun shape in the batch will be used
as the target shape for padding. Otherwise, `padding_to` will be the
target shape (only refers to the second axis).
"""
self._padding_to = padding_to
self._is_training = is_training
def __call__(self, batch):
new_batch = []
# get target shape
max_length = max([audio.shape[1] for audio, _ in batch])
if self._padding_to != -1:
if self._padding_to < max_length:
raise ValueError("If padding_to is not -1, it should be larger "
"than any instance's shape in the batch")
max_length = self._padding_to
max_text_length = max([len(text) for _, text in batch])
# padding
padded_audios = []
audio_lens = []
texts, text_lens = [], []
for audio, text in batch:
# audio
padded_audio = np.zeros([audio.shape[0], max_length])
padded_audio[:, :audio.shape[1]] = audio
padded_audios.append(padded_audio)
audio_lens.append(audio.shape[1])
# text
padded_text = np.zeros([max_text_length])
if self._is_training:
padded_text[:len(text)] = text #ids
else:
padded_text[:len(text)] = [ord(t) for t in text] # string
texts.append(padded_text)
text_lens.append(len(text))
padded_audios = np.array(padded_audios).astype('float32')
audio_lens = np.array(audio_lens).astype('int64')
texts = np.array(texts).astype('int32')
text_lens = np.array(text_lens).astype('int64')
return padded_audios, texts, audio_lens, text_lens
def create_dataloader(manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config='{}',
max_duration=float('inf'),
min_duration=0.0,
stride_ms=10.0,
window_ms=20.0,
max_freq=None,
specgram_type='linear',
use_dB_normalization=True,
random_seed=0,
keep_transcription_text=False,
is_training=False,
batch_size=1,
num_workers=0,
sortagrad=False,
shuffle_method=None,
dist=False):
dataset = DeepSpeech2Dataset(
manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config=augmentation_config,
max_duration=max_duration,
min_duration=min_duration,
stride_ms=stride_ms,
window_ms=window_ms,
max_freq=max_freq,
specgram_type=specgram_type,
use_dB_normalization=use_dB_normalization,
random_seed=random_seed,
keep_transcription_text=keep_transcription_text)
if dist:
batch_sampler = DeepSpeech2DistributedBatchSampler(
dataset,
batch_size,
num_replicas=None,
rank=None,
shuffle=is_training,
drop_last=is_training,
sortagrad=is_training,
shuffle_method=shuffle_method)
else:
batch_sampler = DeepSpeech2BatchSampler(
dataset,
shuffle=is_training,
batch_size=batch_size,
drop_last=is_training,
sortagrad=is_training,
shuffle_method=shuffle_method)
def padding_batch(batch, padding_to=-1, flatten=False, is_training=True):
"""
Padding audio features with zeros to make them have the same shape (or
a user-defined shape) within one bach.
If ``padding_to`` is -1, the maximun shape in the batch will be used
as the target shape for padding. Otherwise, `padding_to` will be the
target shape (only refers to the second axis).
If `flatten` is True, features will be flatten to 1darray.
"""
new_batch = []
# get target shape
max_length = max([audio.shape[1] for audio, text in batch])
if padding_to != -1:
if padding_to < max_length:
raise ValueError("If padding_to is not -1, it should be larger "
"than any instance's shape in the batch")
max_length = padding_to
max_text_length = max([len(text) for audio, text in batch])
# padding
padded_audios = []
audio_lens = []
texts, text_lens = [], []
for audio, text in batch:
padded_audio = np.zeros([audio.shape[0], max_length])
padded_audio[:, :audio.shape[1]] = audio
if flatten:
padded_audio = padded_audio.flatten()
padded_audios.append(padded_audio)
audio_lens.append(audio.shape[1])
padded_text = np.zeros([max_text_length])
if is_training:
padded_text[:len(text)] = text #ids
else:
padded_text[:len(text)] = [ord(t) for t in text] # string
texts.append(padded_text)
text_lens.append(len(text))
padded_audios = np.array(padded_audios).astype('float32')
audio_lens = np.array(audio_lens).astype('int64')
texts = np.array(texts).astype('int32')
text_lens = np.array(text_lens).astype('int64')
return padded_audios, texts, audio_lens, text_lens
loader = DataLoader(
dataset,
batch_sampler=batch_sampler,
collate_fn=partial(padding_batch, is_training=is_training),
num_workers=num_workers)
return loader

@ -27,95 +27,28 @@ import paddle
from paddle import distributed as dist
from paddle.io import DataLoader
from paddle.fluid.dygraph import base as imperative_base
from paddle.fluid import layers
from paddle.fluid import core
from deepspeech.training import Trainer
from deepspeech.utils import mp_tools
from deepspeech.utils.error_rate import char_errors, word_errors, cer, wer
from deepspeech.training.gradclip import MyClipGradByGlobalNorm
from deepspeech.models.network import DeepSpeech2
from deepspeech.models.network import DeepSpeech2Loss
from deepspeech.utils import mp_tools
from deepspeech.utils.error_rate import char_errors
from deepspeech.utils.error_rate import word_errors
from deepspeech.utils.error_rate import cer
from deepspeech.utils.error_rate import wer
from deepspeech.utils.utility import print_grads
from deepspeech.utils.utility import print_params
from deepspeech.decoders.swig_wrapper import Scorer
from deepspeech.decoders.swig_wrapper import ctc_greedy_decoder
from deepspeech.decoders.swig_wrapper import ctc_beam_search_decoder_batch
from deepspeech.io.collator import SpeechCollator
from deepspeech.io.sampler import SortagradDistributedBatchSampler
from deepspeech.io.sampler import SortagradBatchSampler
from deepspeech.io.dataset import ManifestDataset
from deepspeech.exps.deepspeech2.dataset import SpeechCollator
from deepspeech.exps.deepspeech2.dataset import DeepSpeech2Dataset
from deepspeech.exps.deepspeech2.dataset import DeepSpeech2DistributedBatchSampler
from deepspeech.exps.deepspeech2.dataset import DeepSpeech2BatchSampler
from deepspeech.training.loss import CTCLoss
from deepspeech.models.DeepSpeech2 import DeepSpeech2Model
logger = logging.getLogger(__name__)
class MyClipGradByGlobalNorm(paddle.nn.ClipGradByGlobalNorm):
def __init__(self, clip_norm):
super().__init__(clip_norm)
@imperative_base.no_grad
def _dygraph_clip(self, params_grads):
params_and_grads = []
sum_square_list = []
for p, g in params_grads:
if g is None:
continue
if getattr(p, 'need_clip', True) is False:
continue
merge_grad = g
if g.type == core.VarDesc.VarType.SELECTED_ROWS:
merge_grad = layers.merge_selected_rows(g)
merge_grad = layers.get_tensor_from_selected_rows(merge_grad)
square = layers.square(merge_grad)
sum_square = layers.reduce_sum(square)
logger.info(
f"Grad Before Clip: {p.name}: {float(layers.sqrt(layers.reduce_sum(layers.square(merge_grad))) ) }"
)
sum_square_list.append(sum_square)
# all parameters have been filterd out
if len(sum_square_list) == 0:
return params_grads
global_norm_var = layers.concat(sum_square_list)
global_norm_var = layers.reduce_sum(global_norm_var)
global_norm_var = layers.sqrt(global_norm_var)
logger.info(f"Grad Global Norm: {float(global_norm_var)}!!!!")
max_global_norm = layers.fill_constant(
shape=[1], dtype=global_norm_var.dtype, value=self.clip_norm)
clip_var = layers.elementwise_div(
x=max_global_norm,
y=layers.elementwise_max(x=global_norm_var, y=max_global_norm))
for p, g in params_grads:
if g is None:
continue
if getattr(p, 'need_clip', True) is False:
params_and_grads.append((p, g))
continue
new_grad = layers.elementwise_mul(x=g, y=clip_var)
logger.info(
f"Grad After Clip: {p.name}: {float(layers.sqrt(layers.reduce_sum(layers.square(merge_grad))) ) }"
)
params_and_grads.append((p, new_grad))
return params_and_grads
def print_grads(model, logger=None):
for n, p in model.named_parameters():
msg = f"param grad: {n}: shape: {p.shape} grad: {p.grad}"
if logger:
logger.info(msg)
def print_params(model, logger=None):
for n, p in model.named_parameters():
msg = f"param: {n}: shape: {p.shape} stop_grad: {p.stop_gradient}"
if logger:
logger.info(msg)
class DeepSpeech2Trainer(Trainer):
def __init__(self, config, args):
super().__init__(config, args)
@ -193,7 +126,7 @@ class DeepSpeech2Trainer(Trainer):
def setup_model(self):
config = self.config
model = DeepSpeech2(
model = DeepSpeech2Model(
feat_size=self.train_loader.dataset.feature_size,
dict_size=self.train_loader.dataset.vocab_size,
num_conv_layers=config.model.num_conv_layers,
@ -219,7 +152,7 @@ class DeepSpeech2Trainer(Trainer):
config.training.weight_decay),
grad_clip=grad_clip)
criterion = DeepSpeech2Loss(self.train_loader.dataset.vocab_size)
criterion = CTCLoss(self.train_loader.dataset.vocab_size)
self.model = model
self.optimizer = optimizer
@ -230,7 +163,7 @@ class DeepSpeech2Trainer(Trainer):
def setup_dataloader(self):
config = self.config
train_dataset = DeepSpeech2Dataset(
train_dataset = ManifestDataset(
config.data.train_manifest,
config.data.vocab_filepath,
config.data.mean_std_filepath,
@ -250,7 +183,7 @@ class DeepSpeech2Trainer(Trainer):
random_seed=config.data.random_seed,
keep_transcription_text=False)
dev_dataset = DeepSpeech2Dataset(
dev_dataset = ManifestDataset(
config.data.dev_manifest,
config.data.vocab_filepath,
config.data.mean_std_filepath,
@ -269,7 +202,7 @@ class DeepSpeech2Trainer(Trainer):
keep_transcription_text=False)
if self.parallel:
batch_sampler = DeepSpeech2DistributedBatchSampler(
batch_sampler = SortagradDistributedBatchSampler(
train_dataset,
batch_size=config.data.batch_size,
num_replicas=None,
@ -279,7 +212,7 @@ class DeepSpeech2Trainer(Trainer):
sortagrad=config.data.sortagrad,
shuffle_method=config.data.shuffle_method)
else:
batch_sampler = DeepSpeech2BatchSampler(
batch_sampler = SortagradBatchSampler(
train_dataset,
shuffle=True,
batch_size=config.data.batch_size,
@ -461,7 +394,7 @@ class DeepSpeech2Tester(DeepSpeech2Trainer):
def setup_model(self):
config = self.config
model = DeepSpeech2(
model = DeepSpeech2Model(
feat_size=self.test_loader.dataset.feature_size,
dict_size=self.test_loader.dataset.vocab_size,
num_conv_layers=config.model.num_conv_layers,
@ -473,7 +406,7 @@ class DeepSpeech2Tester(DeepSpeech2Trainer):
if self.parallel:
model = paddle.DataParallel(model)
criterion = DeepSpeech2Loss(self.test_loader.dataset.vocab_size)
criterion = CTCLoss(self.test_loader.dataset.vocab_size)
self.model = model
self.criterion = criterion
@ -482,7 +415,7 @@ class DeepSpeech2Tester(DeepSpeech2Trainer):
def setup_dataloader(self):
config = self.config
# return raw text
test_dataset = DeepSpeech2Dataset(
test_dataset = ManifestDataset(
config.data.test_manifest,
config.data.vocab_filepath,
config.data.mean_std_filepath,

@ -0,0 +1,128 @@
# 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.
from paddle.io import DataLoader
from deepspeech.io.collator import SpeechCollator
from deepspeech.io.sampler import SortagradDistributedBatchSampler
from deepspeech.io.sampler import SortagradBatchSampler
from deepspeech.io.dataset import ManifestDataset
def create_dataloader(manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config='{}',
max_duration=float('inf'),
min_duration=0.0,
stride_ms=10.0,
window_ms=20.0,
max_freq=None,
specgram_type='linear',
use_dB_normalization=True,
random_seed=0,
keep_transcription_text=False,
is_training=False,
batch_size=1,
num_workers=0,
sortagrad=False,
shuffle_method=None,
dist=False):
dataset = ManifestDataset(
manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config=augmentation_config,
max_duration=max_duration,
min_duration=min_duration,
stride_ms=stride_ms,
window_ms=window_ms,
max_freq=max_freq,
specgram_type=specgram_type,
use_dB_normalization=use_dB_normalization,
random_seed=random_seed,
keep_transcription_text=keep_transcription_text)
if dist:
batch_sampler = SortagradDistributedBatchSampler(
dataset,
batch_size,
num_replicas=None,
rank=None,
shuffle=is_training,
drop_last=is_training,
sortagrad=is_training,
shuffle_method=shuffle_method)
else:
batch_sampler = SortagradBatchSampler(
dataset,
shuffle=is_training,
batch_size=batch_size,
drop_last=is_training,
sortagrad=is_training,
shuffle_method=shuffle_method)
def padding_batch(batch, padding_to=-1, flatten=False, is_training=True):
"""
Padding audio features with zeros to make them have the same shape (or
a user-defined shape) within one bach.
If ``padding_to`` is -1, the maximun shape in the batch will be used
as the target shape for padding. Otherwise, `padding_to` will be the
target shape (only refers to the second axis).
If `flatten` is True, features will be flatten to 1darray.
"""
new_batch = []
# get target shape
max_length = max([audio.shape[1] for audio, text in batch])
if padding_to != -1:
if padding_to < max_length:
raise ValueError("If padding_to is not -1, it should be larger "
"than any instance's shape in the batch")
max_length = padding_to
max_text_length = max([len(text) for audio, text in batch])
# padding
padded_audios = []
audio_lens = []
texts, text_lens = [], []
for audio, text in batch:
padded_audio = np.zeros([audio.shape[0], max_length])
padded_audio[:, :audio.shape[1]] = audio
if flatten:
padded_audio = padded_audio.flatten()
padded_audios.append(padded_audio)
audio_lens.append(audio.shape[1])
padded_text = np.zeros([max_text_length])
if is_training:
padded_text[:len(text)] = text #ids
else:
padded_text[:len(text)] = [ord(t) for t in text] # string
texts.append(padded_text)
text_lens.append(len(text))
padded_audios = np.array(padded_audios).astype('float32')
audio_lens = np.array(audio_lens).astype('int64')
texts = np.array(texts).astype('int32')
text_lens = np.array(text_lens).astype('int64')
return padded_audios, texts, audio_lens, text_lens
loader = DataLoader(
dataset,
batch_sampler=batch_sampler,
collate_fn=partial(padding_batch, is_training=is_training),
num_workers=num_workers)
return loader

@ -0,0 +1,72 @@
# 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.
import logging
import numpy as np
from collections import namedtuple
logger = logging.getLogger(__name__)
__all__ = [
"SpeechCollator",
]
class SpeechCollator():
def __init__(self, padding_to=-1, is_training=True):
"""
Padding audio features with zeros to make them have the same shape (or
a user-defined shape) within one bach.
If ``padding_to`` is -1, the maximun shape in the batch will be used
as the target shape for padding. Otherwise, `padding_to` will be the
target shape (only refers to the second axis).
"""
self._padding_to = padding_to
self._is_training = is_training
def __call__(self, batch):
new_batch = []
# get target shape
max_length = max([audio.shape[1] for audio, _ in batch])
if self._padding_to != -1:
if self._padding_to < max_length:
raise ValueError("If padding_to is not -1, it should be larger "
"than any instance's shape in the batch")
max_length = self._padding_to
max_text_length = max([len(text) for _, text in batch])
# padding
padded_audios = []
audio_lens = []
texts, text_lens = [], []
for audio, text in batch:
# audio
padded_audio = np.zeros([audio.shape[0], max_length])
padded_audio[:, :audio.shape[1]] = audio
padded_audios.append(padded_audio)
audio_lens.append(audio.shape[1])
# text
padded_text = np.zeros([max_text_length])
if self._is_training:
padded_text[:len(text)] = text #ids
else:
padded_text[:len(text)] = [ord(t) for t in text] # string
texts.append(padded_text)
text_lens.append(len(text))
padded_audios = np.array(padded_audios).astype('float32')
audio_lens = np.array(audio_lens).astype('int64')
texts = np.array(texts).astype('int32')
text_lens = np.array(text_lens).astype('int64')
return padded_audios, texts, audio_lens, text_lens

@ -0,0 +1,186 @@
# 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.
import math
import random
import tarfile
import logging
import numpy as np
from collections import namedtuple
from functools import partial
from paddle.io import Dataset
from deepspeech.frontend.utility import read_manifest
from deepspeech.frontend.augmentor.augmentation import AugmentationPipeline
from deepspeech.frontend.featurizer.speech_featurizer import SpeechFeaturizer
from deepspeech.frontend.speech import SpeechSegment
from deepspeech.frontend.normalizer import FeatureNormalizer
logger = logging.getLogger(__name__)
__all__ = [
"ManifestDataset",
]
class ManifestDataset(Dataset):
def __init__(self,
manifest_path,
vocab_filepath,
mean_std_filepath,
augmentation_config='{}',
max_duration=float('inf'),
min_duration=0.0,
stride_ms=10.0,
window_ms=20.0,
n_fft=None,
max_freq=None,
target_sample_rate=16000,
specgram_type='linear',
use_dB_normalization=True,
target_dB=-20,
random_seed=0,
keep_transcription_text=False):
super().__init__()
self._max_duration = max_duration
self._min_duration = min_duration
self._normalizer = FeatureNormalizer(mean_std_filepath)
self._augmentation_pipeline = AugmentationPipeline(
augmentation_config=augmentation_config, random_seed=random_seed)
self._speech_featurizer = SpeechFeaturizer(
vocab_filepath=vocab_filepath,
specgram_type=specgram_type,
stride_ms=stride_ms,
window_ms=window_ms,
n_fft=n_fft,
max_freq=max_freq,
target_sample_rate=target_sample_rate,
use_dB_normalization=use_dB_normalization,
target_dB=target_dB)
self._rng = random.Random(random_seed)
self._keep_transcription_text = keep_transcription_text
# for caching tar files info
self._local_data = namedtuple('local_data', ['tar2info', 'tar2object'])
self._local_data.tar2info = {}
self._local_data.tar2object = {}
# read manifest
self._manifest = read_manifest(
manifest_path=manifest_path,
max_duration=self._max_duration,
min_duration=self._min_duration)
self._manifest.sort(key=lambda x: x["duration"])
@property
def manifest(self):
return self._manifest
@property
def vocab_size(self):
"""Return the vocabulary size.
:return: Vocabulary size.
:rtype: int
"""
return self._speech_featurizer.vocab_size
@property
def vocab_list(self):
"""Return the vocabulary in list.
:return: Vocabulary in list.
:rtype: list
"""
return self._speech_featurizer.vocab_list
@property
def feature_size(self):
return self._speech_featurizer.feature_size
def _parse_tar(self, file):
"""Parse a tar file to get a tarfile object
and a map containing tarinfoes
"""
result = {}
f = tarfile.open(file)
for tarinfo in f.getmembers():
result[tarinfo.name] = tarinfo
return f, result
def _subfile_from_tar(self, file):
"""Get subfile object from tar.
It will return a subfile object from tar file
and cached tar file info for next reading request.
"""
tarpath, filename = file.split(':', 1)[1].split('#', 1)
if 'tar2info' not in self._local_data.__dict__:
self._local_data.tar2info = {}
if 'tar2object' not in self._local_data.__dict__:
self._local_data.tar2object = {}
if tarpath not in self._local_data.tar2info:
object, infoes = self._parse_tar(tarpath)
self._local_data.tar2info[tarpath] = infoes
self._local_data.tar2object[tarpath] = object
return self._local_data.tar2object[tarpath].extractfile(
self._local_data.tar2info[tarpath][filename])
def process_utterance(self, audio_file, transcript):
"""Load, augment, featurize and normalize for speech data.
:param audio_file: Filepath or file object of audio file.
:type audio_file: str | file
:param transcript: Transcription text.
:type transcript: str
:return: Tuple of audio feature tensor and data of transcription part,
where transcription part could be token ids or text.
:rtype: tuple of (2darray, list)
"""
if isinstance(audio_file, str) and audio_file.startswith('tar:'):
speech_segment = SpeechSegment.from_file(
self._subfile_from_tar(audio_file), transcript)
else:
speech_segment = SpeechSegment.from_file(audio_file, transcript)
self._augmentation_pipeline.transform_audio(speech_segment)
specgram, transcript_part = self._speech_featurizer.featurize(
speech_segment, self._keep_transcription_text)
specgram = self._normalizer.apply(specgram)
return specgram, transcript_part
def _instance_reader_creator(self, manifest):
"""
Instance reader creator. Create a callable function to produce
instances of data.
Instance: a tuple of ndarray of audio spectrogram and a list of
token indices for transcript.
"""
def reader():
for instance in manifest:
inst = self.process_utterance(instance["audio_filepath"],
instance["text"])
yield inst
return reader
def __len__(self):
return len(self._manifest)
def __getitem__(self, idx):
instance = self._manifest[idx]
return self.process_utterance(instance["audio_filepath"],
instance["text"])

@ -0,0 +1,266 @@
# 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.
import math
import random
import tarfile
import logging
import numpy as np
from collections import namedtuple
from functools import partial
import paddle
from paddle.io import BatchSampler
from paddle.io import DistributedBatchSampler
from paddle import distributed as dist
logger = logging.getLogger(__name__)
__all__ = [
"SortagradDistributedBatchSampler",
"SortagradBatchSampler",
]
class SortagradDistributedBatchSampler(DistributedBatchSampler):
def __init__(self,
dataset,
batch_size,
num_replicas=None,
rank=None,
shuffle=False,
drop_last=False,
sortagrad=False,
shuffle_method="batch_shuffle"):
super().__init__(dataset, batch_size, num_replicas, rank, shuffle,
drop_last)
self._sortagrad = sortagrad
self._shuffle_method = shuffle_method
def _batch_shuffle(self, indices, batch_size, clipped=False):
"""Put similarly-sized instances into minibatches for better efficiency
and make a batch-wise shuffle.
1. Sort the audio clips by duration.
2. Generate a random number `k`, k in [0, batch_size).
3. Randomly shift `k` instances in order to create different batches
for different epochs. Create minibatches.
4. Shuffle the minibatches.
:param indices: indexes. List of int.
:type indices: list
:param batch_size: Batch size. This size is also used for generate
a random number for batch shuffle.
:type batch_size: int
:param clipped: Whether to clip the heading (small shift) and trailing
(incomplete batch) instances.
:type clipped: bool
:return: Batch shuffled mainifest.
:rtype: list
"""
rng = np.random.RandomState(self.epoch)
shift_len = rng.randint(0, batch_size - 1)
batch_indices = list(zip(* [iter(indices[shift_len:])] * batch_size))
rng.shuffle(batch_indices)
batch_indices = [item for batch in batch_indices for item in batch]
assert (clipped == False)
if not clipped:
res_len = len(indices) - shift_len - len(batch_indices)
# when res_len is 0, will return whole list, len(List[-0:]) = len(List[:])
if res_len != 0:
batch_indices.extend(indices[-res_len:])
batch_indices.extend(indices[0:shift_len])
assert len(indices) == len(
batch_indices
), f"_batch_shuffle: {len(indices)} : {len(batch_indices)} : {res_len} - {shift_len}"
return batch_indices
def __iter__(self):
num_samples = len(self.dataset)
indices = np.arange(num_samples).tolist()
indices += indices[:(self.total_size - len(indices))]
assert len(indices) == self.total_size
# sort (by duration) or batch-wise shuffle the manifest
if self.shuffle:
if self.epoch == 0 and self._sortagrad:
logger.info(
f'rank: {dist.get_rank()} dataset sortagrad! epoch {self.epoch}'
)
else:
logger.info(
f'rank: {dist.get_rank()} dataset shuffle! epoch {self.epoch}'
)
if self._shuffle_method == "batch_shuffle":
indices = self._batch_shuffle(
indices, self.batch_size, clipped=False)
elif self._shuffle_method == "instance_shuffle":
np.random.RandomState(self.epoch).shuffle(indices)
else:
raise ValueError("Unknown shuffle method %s." %
self._shuffle_method)
assert len(
indices
) == self.total_size, f"batch shuffle examples error: {len(indices)} : {self.total_size}"
# subsample
def _get_indices_by_batch_size(indices):
subsampled_indices = []
last_batch_size = self.total_size % (self.batch_size * self.nranks)
assert last_batch_size % self.nranks == 0
last_local_batch_size = last_batch_size // self.nranks
for i in range(self.local_rank * self.batch_size,
len(indices) - last_batch_size,
self.batch_size * self.nranks):
subsampled_indices.extend(indices[i:i + self.batch_size])
indices = indices[len(indices) - last_batch_size:]
subsampled_indices.extend(
indices[self.local_rank * last_local_batch_size:(
self.local_rank + 1) * last_local_batch_size])
return subsampled_indices
if self.nranks > 1:
indices = _get_indices_by_batch_size(indices)
assert len(indices) == self.num_samples
_sample_iter = iter(indices)
batch_indices = []
for idx in _sample_iter:
batch_indices.append(idx)
if len(batch_indices) == self.batch_size:
logger.info(
f"rank: {dist.get_rank()} batch index: {batch_indices} ")
yield batch_indices
batch_indices = []
if not self.drop_last and len(batch_indices) > 0:
yield batch_indices
def __len__(self):
num_samples = self.num_samples
num_samples += int(not self.drop_last) * (self.batch_size - 1)
return num_samples // self.batch_size
class SortagradBatchSampler(BatchSampler):
def __init__(self,
dataset,
batch_size,
shuffle=False,
drop_last=False,
sortagrad=False,
shuffle_method="batch_shuffle"):
self.dataset = dataset
assert isinstance(batch_size, int) and batch_size > 0, \
"batch_size should be a positive integer"
self.batch_size = batch_size
assert isinstance(shuffle, bool), \
"shuffle should be a boolean value"
self.shuffle = shuffle
assert isinstance(drop_last, bool), \
"drop_last should be a boolean number"
self.drop_last = drop_last
self.epoch = 0
self.num_samples = int(math.ceil(len(self.dataset) * 1.0))
self.total_size = self.num_samples
self._sortagrad = sortagrad
self._shuffle_method = shuffle_method
def _batch_shuffle(self, indices, batch_size, clipped=False):
"""Put similarly-sized instances into minibatches for better efficiency
and make a batch-wise shuffle.
1. Sort the audio clips by duration.
2. Generate a random number `k`, k in [0, batch_size).
3. Randomly shift `k` instances in order to create different batches
for different epochs. Create minibatches.
4. Shuffle the minibatches.
:param indices: indexes. List of int.
:type indices: list
:param batch_size: Batch size. This size is also used for generate
a random number for batch shuffle.
:type batch_size: int
:param clipped: Whether to clip the heading (small shift) and trailing
(incomplete batch) instances.
:type clipped: bool
:return: Batch shuffled mainifest.
:rtype: list
"""
rng = np.random.RandomState(self.epoch)
# must shift at leat by one
shift_len = rng.randint(0, batch_size - 1)
batch_indices = list(zip(* [iter(indices[shift_len:])] * batch_size))
rng.shuffle(batch_indices)
batch_indices = [item for batch in batch_indices for item in batch]
assert (clipped == False)
if not clipped:
res_len = len(indices) - shift_len - len(batch_indices)
# when res_len is 0, will return whole list, len(List[-0:]) = len(List[:])
if res_len != 0:
batch_indices.extend(indices[-res_len:])
batch_indices.extend(indices[0:shift_len])
assert len(indices) == len(
batch_indices
), f"_batch_shuffle: {len(indices)} : {len(batch_indices)} : {res_len} - {shift_len}"
return batch_indices
def __iter__(self):
num_samples = len(self.dataset)
indices = np.arange(num_samples).tolist()
indices += indices[:(self.total_size - len(indices))]
assert len(indices) == self.total_size
# sort (by duration) or batch-wise shuffle the manifest
if self.shuffle:
if self.epoch == 0 and self._sortagrad:
logger.info(f'dataset sortagrad! epoch {self.epoch}')
else:
logger.info(f'dataset shuffle! epoch {self.epoch}')
if self._shuffle_method == "batch_shuffle":
indices = self._batch_shuffle(
indices, self.batch_size, clipped=False)
elif self._shuffle_method == "instance_shuffle":
np.random.RandomState(self.epoch).shuffle(indices)
else:
raise ValueError("Unknown shuffle method %s." %
self._shuffle_method)
assert len(
indices
) == self.total_size, f"batch shuffle examples error: {len(indices)} : {self.total_size}"
assert len(indices) == self.num_samples
_sample_iter = iter(indices)
batch_indices = []
for idx in _sample_iter:
batch_indices.append(idx)
if len(batch_indices) == self.batch_size:
logger.info(
f"rank: {dist.get_rank()} batch index: {batch_indices} ")
yield batch_indices
batch_indices = []
if not self.drop_last and len(batch_indices) > 0:
yield batch_indices
self.epoch += 1
def __len__(self):
num_samples = self.num_samples
num_samples += int(not self.drop_last) * (self.batch_size - 1)
return num_samples // self.batch_size

@ -0,0 +1,304 @@
# 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.
import math
import collections
import numpy as np
import logging
from typing import Optional
from yacs.config import CfgNode
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
from deepspeech.modules.conv import ConvStack
from deepspeech.modules.conv import RNNStack
from deepspeech.modules.mask import sequence_mask
from deepspeech.modules.activation import brelu
from deepspeech.utils import checkpoint
from deepspeech.decoders.swig_wrapper import Scorer
from deepspeech.decoders.swig_wrapper import ctc_greedy_decoder
from deepspeech.decoders.swig_wrapper import ctc_beam_search_decoder_batch
logger = logging.getLogger(__name__)
__all__ = ['DeepSpeech2Model']
class DeepSpeech2Model(nn.Layer):
"""The DeepSpeech2 network structure.
:param audio_data: Audio spectrogram data layer.
:type audio_data: Variable
:param text_data: Transcription text data layer.
:type text_data: Variable
:param audio_len: Valid sequence length data layer.
:type audio_len: Variable
:param masks: Masks data layer to reset padding.
:type masks: Variable
:param dict_size: Dictionary size for tokenized transcription.
:type dict_size: int
:param num_conv_layers: Number of stacking convolution layers.
:type num_conv_layers: int
:param num_rnn_layers: Number of stacking RNN layers.
:type num_rnn_layers: int
:param rnn_size: RNN layer size (dimension of RNN cells).
:type rnn_size: int
:param use_gru: Use gru if set True. Use simple rnn if set False.
:type use_gru: bool
:param share_rnn_weights: Whether to share input-hidden weights between
forward and backward direction RNNs.
It is only available when use_gru=False.
:type share_weights: bool
:return: A tuple of an output unnormalized log probability layer (
before softmax) and a ctc cost layer.
:rtype: tuple of LayerOutput
"""
@classmethod
def params(cls, config: Optional[CfgNode]=None) -> CfgNode:
default = CfgNode(
dict(
num_conv_layers=2, #Number of stacking convolution layers.
num_rnn_layers=3, #Number of stacking RNN layers.
rnn_layer_size=1024, #RNN layer size (number of RNN cells).
use_gru=True, #Use gru if set True. Use simple rnn if set False.
share_rnn_weights=True #Whether to share input-hidden weights between forward and backward directional RNNs.Notice that for GRU, weight sharing is not supported.
))
if config is not None:
config.model.merge_from_other_cfg(default)
return default
def __init__(self,
feat_size,
dict_size,
num_conv_layers=2,
num_rnn_layers=3,
rnn_size=1024,
use_gru=False,
share_rnn_weights=True):
super().__init__()
self.feat_size = feat_size # 161 for linear
self.dict_size = dict_size
self.conv = ConvStack(feat_size, num_conv_layers)
i_size = self.conv.output_height # H after conv stack
self.rnn = RNNStack(
i_size=i_size,
h_size=rnn_size,
num_stacks=num_rnn_layers,
use_gru=use_gru,
share_rnn_weights=share_rnn_weights)
self.fc = nn.Linear(rnn_size * 2, dict_size + 1)
self.logger = logging.getLogger(__name__)
self._ext_scorer = None
def infer(self, audio, audio_len):
# [B, D, T] -> [B, C=1, D, T]
audio = audio.unsqueeze(1)
# convolution group
x, audio_len = self.conv(audio, audio_len)
#print('conv out', x.shape)
# convert data from convolution feature map to sequence of vectors
B, C, D, T = paddle.shape(x)
x = x.transpose([0, 3, 1, 2]) #[B, T, C, D]
x = x.reshape([B, T, C * D]) #[B, T, C*D]
#print('rnn input', x.shape)
# remove padding part
x, audio_len = self.rnn(x, audio_len) #[B, T, D]
#print('rnn output', x.shape)
logits = self.fc(x) #[B, T, V + 1]
#ctcdecoder need probs, not log_probs
probs = F.softmax(logits)
return logits, probs, audio_len
def forward(self, audio, text, audio_len, text_len):
"""
audio: shape [B, D, T]
text: shape [B, T]
audio_len: shape [B]
text_len: shape [B]
"""
return self.infer(audio, audio_len)
@paddle.no_grad()
def predict(self, audio, audio_len):
""" Model infer """
return self.infer(audio, audio_len)
def _decode_batch_greedy(self, probs_split, vocab_list):
"""Decode by best path for a batch of probs matrix input.
:param probs_split: List of 2-D probability matrix, and each consists
of prob vectors for one speech utterancce.
:param probs_split: List of matrix
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
:return: List of transcription texts.
:rtype: List of str
"""
results = []
for i, probs in enumerate(probs_split):
output_transcription = ctc_greedy_decoder(
probs_seq=probs, vocabulary=vocab_list)
results.append(output_transcription)
return results
def _init_ext_scorer(self, beam_alpha, beam_beta, language_model_path,
vocab_list):
"""Initialize the external scorer.
:param beam_alpha: Parameter associated with language model.
:type beam_alpha: float
:param beam_beta: Parameter associated with word count.
:type beam_beta: float
:param language_model_path: Filepath for language model. If it is
empty, the external scorer will be set to
None, and the decoding method will be pure
beam search without scorer.
:type language_model_path: str|None
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
"""
# init once
if self._ext_scorer != None:
return
if language_model_path != '':
self.logger.info("begin to initialize the external scorer "
"for decoding")
self._ext_scorer = Scorer(beam_alpha, beam_beta,
language_model_path, vocab_list)
lm_char_based = self._ext_scorer.is_character_based()
lm_max_order = self._ext_scorer.get_max_order()
lm_dict_size = self._ext_scorer.get_dict_size()
self.logger.info("language model: "
"is_character_based = %d," % lm_char_based +
" max_order = %d," % lm_max_order +
" dict_size = %d" % lm_dict_size)
self.logger.info("end initializing scorer")
else:
self._ext_scorer = None
self.logger.info("no language model provided, "
"decoding by pure beam search without scorer.")
def _decode_batch_beam_search(self, probs_split, beam_alpha, beam_beta,
beam_size, cutoff_prob, cutoff_top_n,
vocab_list, num_processes):
"""Decode by beam search for a batch of probs matrix input.
:param probs_split: List of 2-D probability matrix, and each consists
of prob vectors for one speech utterancce.
:param probs_split: List of matrix
:param beam_alpha: Parameter associated with language model.
:type beam_alpha: float
:param beam_beta: Parameter associated with word count.
:type beam_beta: float
:param beam_size: Width for Beam search.
:type beam_size: int
:param cutoff_prob: Cutoff probability in pruning,
default 1.0, no pruning.
:type cutoff_prob: float
:param cutoff_top_n: Cutoff number in pruning, only top cutoff_top_n
characters with highest probs in vocabulary will be
used in beam search, default 40.
:type cutoff_top_n: int
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
:param num_processes: Number of processes (CPU) for decoder.
:type num_processes: int
:return: List of transcription texts.
:rtype: List of str
"""
if self._ext_scorer != None:
self._ext_scorer.reset_params(beam_alpha, beam_beta)
# beam search decode
num_processes = min(num_processes, len(probs_split))
beam_search_results = ctc_beam_search_decoder_batch(
probs_split=probs_split,
vocabulary=vocab_list,
beam_size=beam_size,
num_processes=num_processes,
ext_scoring_func=self._ext_scorer,
cutoff_prob=cutoff_prob,
cutoff_top_n=cutoff_top_n)
results = [result[0][1] for result in beam_search_results]
return results
def init_decode(self, beam_alpha, beam_beta, lang_model_path, vocab_list,
decoding_method):
if decoding_method == "ctc_beam_search":
self._init_ext_scorer(beam_alpha, beam_beta, lang_model_path,
vocab_list)
def decode_probs(self, probs, logits_lens, vocab_list, decoding_method,
lang_model_path, beam_alpha, beam_beta, beam_size,
cutoff_prob, cutoff_top_n, num_processes):
""" probs: activation after softmax
logits_len: audio output lens
"""
probs_split = [probs[i, :l, :] for i, l in enumerate(logits_lens)]
if decoding_method == "ctc_greedy":
result_transcripts = self._decode_batch_greedy(
probs_split=probs_split, vocab_list=vocab_list)
elif decoding_method == "ctc_beam_search":
result_transcripts = self._decode_batch_beam_search(
probs_split=probs_split,
beam_alpha=beam_alpha,
beam_beta=beam_beta,
beam_size=beam_size,
cutoff_prob=cutoff_prob,
cutoff_top_n=cutoff_top_n,
vocab_list=vocab_list,
num_processes=num_processes)
else:
raise ValueError(f"Not support: {decoding_method}")
return result_transcripts
@paddle.no_grad()
def decode(self, audio, audio_len, vocab_list, decoding_method,
lang_model_path, beam_alpha, beam_beta, beam_size, cutoff_prob,
cutoff_top_n, num_processes):
_, probs, logits_lens = self.predict(audio, audio_len)
return self.decode_probs(probs.numpy(), logits_lens, vocab_list,
decoding_method, lang_model_path, beam_alpha,
beam_beta, beam_size, cutoff_prob,
cutoff_top_n, num_processes)
def from_pretrained(self, checkpoint_path):
"""Build a model from a pretrained model.
Parameters
----------
model: nn.Layer
Asr Model.
checkpoint_path: Path or str
The path of pretrained model checkpoint, without extension name.
Returns
-------
Model
The model build from pretrined result.
"""
checkpoint.load_parameters(self, checkpoint_path=checkpoint_path)
return

@ -1,754 +0,0 @@
# 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.
import math
import collections
import numpy as np
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
from deepspeech.utils import checkpoint
from deepspeech.decoders.swig_wrapper import Scorer
from deepspeech.decoders.swig_wrapper import ctc_greedy_decoder
from deepspeech.decoders.swig_wrapper import ctc_beam_search_decoder_batch
logger = logging.getLogger(__name__)
__all__ = ['DeepSpeech2', 'DeepSpeech2Loss']
def brelu(x, t_min=0.0, t_max=24.0, name=None):
t_min = paddle.to_tensor(t_min)
t_max = paddle.to_tensor(t_max)
return x.maximum(t_min).minimum(t_max)
def sequence_mask(x_len, max_len=None, dtype='float32'):
max_len = max_len or x_len.max()
x_len = paddle.unsqueeze(x_len, -1)
row_vector = paddle.arange(max_len)
#mask = row_vector < x_len
mask = row_vector > x_len # a bug, broadcast 的时候出错了
mask = paddle.cast(mask, dtype)
return mask
class ConvBn(nn.Layer):
"""Convolution layer with batch normalization.
:param kernel_size: The x dimension of a filter kernel. Or input a tuple for
two image dimension.
:type kernel_size: int|tuple|list
:param num_channels_in: Number of input channels.
:type num_channels_in: int
:param num_channels_out: Number of output channels.
:type num_channels_out: int
:param stride: The x dimension of the stride. Or input a tuple for two
image dimension.
:type stride: int|tuple|list
:param padding: The x dimension of the padding. Or input a tuple for two
image dimension.
:type padding: int|tuple|list
:param act: Activation type, relu|brelu
:type act: string
:param masks: Masks data layer to reset padding.
:type masks: Variable
:param name: Name of the layer.
:param name: string
:return: Batch norm layer after convolution layer.
:rtype: Variable
"""
def __init__(self, num_channels_in, num_channels_out, kernel_size, stride,
padding, act):
super().__init__()
self.kernel_size = kernel_size
self.stride = stride
self.padding = padding
self.conv = nn.Conv2D(
num_channels_in,
num_channels_out,
kernel_size=kernel_size,
stride=stride,
padding=padding,
weight_attr=None,
bias_attr=False,
data_format='NCHW')
self.bn = nn.BatchNorm2D(
num_channels_out,
weight_attr=None,
bias_attr=None,
data_format='NCHW')
self.act = F.relu if act == 'relu' else brelu
def forward(self, x, x_len):
"""
x(Tensor): audio, shape [B, C, D, T]
"""
x = self.conv(x)
x = self.bn(x)
x = self.act(x)
x_len = (x_len - self.kernel_size[1] + 2 * self.padding[1]
) // self.stride[1] + 1
# reset padding part to 0
masks = sequence_mask(x_len) #[B, T]
masks = masks.unsqueeze(1).unsqueeze(1) # [B, 1, 1, T]
x = x.multiply(masks)
return x, x_len
class ConvStack(nn.Layer):
"""Convolution group with stacked convolution layers.
:param feat_size: audio feature dim.
:type feat_size: int
:param num_stacks: Number of stacked convolution layers.
:type num_stacks: int
"""
def __init__(self, feat_size, num_stacks):
super().__init__()
self.feat_size = feat_size # D
self.num_stacks = num_stacks
self.conv_in = ConvBn(
num_channels_in=1,
num_channels_out=32,
kernel_size=(41, 11), #[D, T]
stride=(2, 3),
padding=(20, 5),
act='brelu')
out_channel = 32
self.conv_stack = nn.LayerList([
ConvBn(
num_channels_in=32,
num_channels_out=out_channel,
kernel_size=(21, 11),
stride=(2, 1),
padding=(10, 5),
act='brelu') for i in range(num_stacks - 1)
])
# conv output feat_dim
output_height = (feat_size - 1) // 2 + 1
for i in range(self.num_stacks - 1):
output_height = (output_height - 1) // 2 + 1
self.output_height = out_channel * output_height
def forward(self, x, x_len):
"""
x: shape [B, C, D, T]
x_len : shape [B]
"""
x, x_len = self.conv_in(x, x_len)
for i, conv in enumerate(self.conv_stack):
x, x_len = conv(x, x_len)
return x, x_len
class RNNCell(nn.RNNCellBase):
r"""
Elman RNN (SimpleRNN) cell. Given the inputs and previous states, it
computes the outputs and updates states.
The formula used is as follows:
.. math::
h_{t} & = act(x_{t} + b_{ih} + W_{hh}h_{t-1} + b_{hh})
y_{t} & = h_{t}
where :math:`act` is for :attr:`activation`.
"""
def __init__(self,
hidden_size,
activation="tanh",
weight_ih_attr=None,
weight_hh_attr=None,
bias_ih_attr=None,
bias_hh_attr=None,
name=None):
super().__init__()
std = 1.0 / math.sqrt(hidden_size)
self.weight_hh = self.create_parameter(
(hidden_size, hidden_size),
weight_hh_attr,
default_initializer=I.Uniform(-std, std))
# self.bias_ih = self.create_parameter(
# (hidden_size, ),
# bias_ih_attr,
# is_bias=True,
# default_initializer=I.Uniform(-std, std))
self.bias_ih = None
self.bias_hh = self.create_parameter(
(hidden_size, ),
bias_hh_attr,
is_bias=True,
default_initializer=I.Uniform(-std, std))
self.hidden_size = hidden_size
if activation not in ["tanh", "relu", "brelu"]:
raise ValueError(
"activation for SimpleRNNCell should be tanh or relu, "
"but get {}".format(activation))
self.activation = activation
self._activation_fn = paddle.tanh \
if activation == "tanh" \
else F.relu
if activation == 'brelu':
self._activation_fn = brelu
def forward(self, inputs, states=None):
if states is None:
states = self.get_initial_states(inputs, self.state_shape)
pre_h = states
i2h = inputs
if self.bias_ih is not None:
i2h += self.bias_ih
h2h = paddle.matmul(pre_h, self.weight_hh, transpose_y=True)
if self.bias_hh is not None:
h2h += self.bias_hh
h = self._activation_fn(i2h + h2h)
return h, h
@property
def state_shape(self):
return (self.hidden_size, )
class GRUCellShare(nn.RNNCellBase):
r"""
Gated Recurrent Unit (GRU) RNN cell. Given the inputs and previous states,
it computes the outputs and updates states.
The formula for GRU used is as follows:
.. math::
r_{t} & = \sigma(W_{ir}x_{t} + b_{ir} + W_{hr}h_{t-1} + b_{hr})
z_{t} & = \sigma(W_{iz}x_{t} + b_{iz} + W_{hz}h_{t-1} + b_{hz})
\widetilde{h}_{t} & = \tanh(W_{ic}x_{t} + b_{ic} + r_{t} * (W_{hc}h_{t-1} + b_{hc}))
h_{t} & = z_{t} * h_{t-1} + (1 - z_{t}) * \widetilde{h}_{t}
y_{t} & = h_{t}
where :math:`\sigma` is the sigmoid fucntion, and * is the elemetwise
multiplication operator.
"""
def __init__(self,
input_size,
hidden_size,
weight_ih_attr=None,
weight_hh_attr=None,
bias_ih_attr=None,
bias_hh_attr=None,
name=None):
super().__init__()
std = 1.0 / math.sqrt(hidden_size)
self.weight_hh = self.create_parameter(
(3 * hidden_size, hidden_size),
weight_hh_attr,
default_initializer=I.Uniform(-std, std))
# self.bias_ih = self.create_parameter(
# (3 * hidden_size, ),
# bias_ih_attr,
# is_bias=True,
# default_initializer=I.Uniform(-std, std))
self.bias_ih = None
self.bias_hh = self.create_parameter(
(3 * hidden_size, ),
bias_hh_attr,
is_bias=True,
default_initializer=I.Uniform(-std, std))
self.hidden_size = hidden_size
self.input_size = input_size
self._gate_activation = F.sigmoid
self._activation = paddle.tanh
#self._activation = F.relu
def forward(self, inputs, states=None):
if states is None:
states = self.get_initial_states(inputs, self.state_shape)
pre_hidden = states
x_gates = inputs
if self.bias_ih is not None:
x_gates = x_gates + self.bias_ih
h_gates = paddle.matmul(pre_hidden, self.weight_hh, transpose_y=True)
if self.bias_hh is not None:
h_gates = h_gates + self.bias_hh
x_r, x_z, x_c = paddle.split(x_gates, num_or_sections=3, axis=1)
h_r, h_z, h_c = paddle.split(h_gates, num_or_sections=3, axis=1)
r = self._gate_activation(x_r + h_r)
z = self._gate_activation(x_z + h_z)
c = self._activation(x_c + r * h_c) # apply reset gate after mm
h = (pre_hidden - c) * z + c
# https://www.paddlepaddle.org.cn/documentation/docs/zh/api/paddle/fluid/layers/dynamic_gru_cn.html#dynamic-gru
return h, h
@property
def state_shape(self):
r"""
The `state_shape` of GRUCell is a shape `[hidden_size]` (-1 for batch
size would be automatically inserted into shape). The shape corresponds
to the shape of :math:`h_{t-1}`.
"""
return (self.hidden_size, )
class BiRNNWithBN(nn.Layer):
"""Bidirectonal simple rnn layer with sequence-wise batch normalization.
The batch normalization is only performed on input-state weights.
:param name: Name of the layer parameters.
:type name: string
:param size: Dimension of RNN cells.
:type size: int
:param share_weights: Whether to share input-hidden weights between
forward and backward directional RNNs.
:type share_weights: bool
:return: Bidirectional simple rnn layer.
:rtype: Variable
"""
def __init__(self, i_size, h_size, share_weights):
super().__init__()
self.share_weights = share_weights
if self.share_weights:
#input-hidden weights shared between bi-directional rnn.
self.fw_fc = nn.Linear(i_size, h_size, bias_attr=False)
# batch norm is only performed on input-state projection
self.fw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.bw_fc = self.fw_fc
self.bw_bn = self.fw_bn
else:
self.fw_fc = nn.Linear(i_size, h_size, bias_attr=False)
self.fw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.bw_fc = nn.Linear(i_size, h_size, bias_attr=False)
self.bw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.fw_cell = RNNCell(hidden_size=h_size, activation='brelu')
self.bw_cell = RNNCell(hidden_size=h_size, activation='brelu')
self.fw_rnn = nn.RNN(
self.fw_cell, is_reverse=False, time_major=False) #[B, T, D]
self.bw_rnn = nn.RNN(
self.fw_cell, is_reverse=True, time_major=False) #[B, T, D]
def forward(self, x, x_len):
# x, shape [B, T, D]
fw_x = self.fw_bn(self.fw_fc(x))
bw_x = self.bw_bn(self.bw_fc(x))
fw_x, _ = self.fw_rnn(inputs=fw_x, sequence_length=x_len)
bw_x, _ = self.bw_rnn(inputs=bw_x, sequence_length=x_len)
x = paddle.concat([fw_x, bw_x], axis=-1)
return x, x_len
class BiGRUWithBN(nn.Layer):
"""Bidirectonal gru layer with sequence-wise batch normalization.
The batch normalization is only performed on input-state weights.
:param name: Name of the layer.
:type name: string
:param input: Input layer.
:type input: Variable
:param size: Dimension of GRU cells.
:type size: int
:param act: Activation type.
:type act: string
:return: Bidirectional GRU layer.
:rtype: Variable
"""
def __init__(self, i_size, h_size, act):
super().__init__()
hidden_size = h_size * 3
self.fw_fc = nn.Linear(i_size, hidden_size, bias_attr=False)
self.fw_bn = nn.BatchNorm1D(
hidden_size, bias_attr=None, data_format='NLC')
self.bw_fc = nn.Linear(i_size, hidden_size, bias_attr=False)
self.bw_bn = nn.BatchNorm1D(
hidden_size, bias_attr=None, data_format='NLC')
self.fw_cell = GRUCellShare(input_size=hidden_size, hidden_size=h_size)
self.bw_cell = GRUCellShare(input_size=hidden_size, hidden_size=h_size)
self.fw_rnn = nn.RNN(
self.fw_cell, is_reverse=False, time_major=False) #[B, T, D]
self.bw_rnn = nn.RNN(
self.fw_cell, is_reverse=True, time_major=False) #[B, T, D]
def forward(self, x, x_len):
# x, shape [B, T, D]
fw_x = self.fw_bn(self.fw_fc(x))
bw_x = self.bw_bn(self.bw_fc(x))
fw_x, _ = self.fw_rnn(inputs=fw_x, sequence_length=x_len)
bw_x, _ = self.bw_rnn(inputs=bw_x, sequence_length=x_len)
x = paddle.concat([fw_x, bw_x], axis=-1)
return x, x_len
class RNNStack(nn.Layer):
"""RNN group with stacked bidirectional simple RNN or GRU layers.
:param input: Input layer.
:type input: Variable
:param size: Dimension of RNN cells in each layer.
:type size: int
:param num_stacks: Number of stacked rnn layers.
:type num_stacks: int
:param use_gru: Use gru if set True. Use simple rnn if set False.
:type use_gru: bool
:param share_rnn_weights: Whether to share input-hidden weights between
forward and backward directional RNNs.
It is only available when use_gru=False.
:type share_weights: bool
:return: Output layer of the RNN group.
:rtype: Variable
"""
def __init__(self, i_size, h_size, num_stacks, use_gru, share_rnn_weights):
super().__init__()
self.rnn_stacks = nn.LayerList()
for i in range(num_stacks):
if use_gru:
#default:GRU using tanh
self.rnn_stacks.append(
BiGRUWithBN(i_size=i_size, h_size=h_size, act="relu"))
else:
self.rnn_stacks.append(
BiRNNWithBN(
i_size=i_size,
h_size=h_size,
share_weights=share_rnn_weights))
i_size = h_size * 2
def forward(self, x, x_len):
"""
x: shape [B, T, D]
x_len: shpae [B]
"""
for i, rnn in enumerate(self.rnn_stacks):
x, x_len = rnn(x, x_len)
masks = sequence_mask(x_len) #[B, T]
masks = masks.unsqueeze(-1) # [B, T, 1]
x = x.multiply(masks)
return x, x_len
class DeepSpeech2(nn.Layer):
"""The DeepSpeech2 network structure.
:param audio_data: Audio spectrogram data layer.
:type audio_data: Variable
:param text_data: Transcription text data layer.
:type text_data: Variable
:param audio_len: Valid sequence length data layer.
:type audio_len: Variable
:param masks: Masks data layer to reset padding.
:type masks: Variable
:param dict_size: Dictionary size for tokenized transcription.
:type dict_size: int
:param num_conv_layers: Number of stacking convolution layers.
:type num_conv_layers: int
:param num_rnn_layers: Number of stacking RNN layers.
:type num_rnn_layers: int
:param rnn_size: RNN layer size (dimension of RNN cells).
:type rnn_size: int
:param use_gru: Use gru if set True. Use simple rnn if set False.
:type use_gru: bool
:param share_rnn_weights: Whether to share input-hidden weights between
forward and backward direction RNNs.
It is only available when use_gru=False.
:type share_weights: bool
:return: A tuple of an output unnormalized log probability layer (
before softmax) and a ctc cost layer.
:rtype: tuple of LayerOutput
"""
def __init__(self,
feat_size,
dict_size,
num_conv_layers=2,
num_rnn_layers=3,
rnn_size=1024,
use_gru=False,
share_rnn_weights=True):
super().__init__()
self.feat_size = feat_size # 161 for linear
self.dict_size = dict_size
self.conv = ConvStack(feat_size, num_conv_layers)
i_size = self.conv.output_height # H after conv stack
self.rnn = RNNStack(
i_size=i_size,
h_size=rnn_size,
num_stacks=num_rnn_layers,
use_gru=use_gru,
share_rnn_weights=share_rnn_weights)
self.fc = nn.Linear(rnn_size * 2, dict_size + 1)
self.logger = logging.getLogger(__name__)
self._ext_scorer = None
def infer(self, audio, audio_len):
# [B, D, T] -> [B, C=1, D, T]
audio = audio.unsqueeze(1)
# convolution group
x, audio_len = self.conv(audio, audio_len)
#print('conv out', x.shape)
# convert data from convolution feature map to sequence of vectors
B, C, D, T = paddle.shape(x)
x = x.transpose([0, 3, 1, 2]) #[B, T, C, D]
x = x.reshape([B, T, C * D]) #[B, T, C*D]
#print('rnn input', x.shape)
# remove padding part
x, audio_len = self.rnn(x, audio_len) #[B, T, D]
#print('rnn output', x.shape)
logits = self.fc(x) #[B, T, V + 1]
#ctcdecoder need probs, not log_probs
probs = F.softmax(logits)
return logits, probs, audio_len
def forward(self, audio, text, audio_len, text_len):
"""
audio: shape [B, D, T]
text: shape [B, T]
audio_len: shape [B]
text_len: shape [B]
"""
return self.infer(audio, audio_len)
@paddle.no_grad()
def predict(self, audio, audio_len):
""" Model infer """
return self.infer(audio, audio_len)
def _decode_batch_greedy(self, probs_split, vocab_list):
"""Decode by best path for a batch of probs matrix input.
:param probs_split: List of 2-D probability matrix, and each consists
of prob vectors for one speech utterancce.
:param probs_split: List of matrix
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
:return: List of transcription texts.
:rtype: List of str
"""
results = []
for i, probs in enumerate(probs_split):
output_transcription = ctc_greedy_decoder(
probs_seq=probs, vocabulary=vocab_list)
results.append(output_transcription)
return results
def _init_ext_scorer(self, beam_alpha, beam_beta, language_model_path,
vocab_list):
"""Initialize the external scorer.
:param beam_alpha: Parameter associated with language model.
:type beam_alpha: float
:param beam_beta: Parameter associated with word count.
:type beam_beta: float
:param language_model_path: Filepath for language model. If it is
empty, the external scorer will be set to
None, and the decoding method will be pure
beam search without scorer.
:type language_model_path: str|None
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
"""
# init once
if self._ext_scorer != None:
return
if language_model_path != '':
self.logger.info("begin to initialize the external scorer "
"for decoding")
self._ext_scorer = Scorer(beam_alpha, beam_beta,
language_model_path, vocab_list)
lm_char_based = self._ext_scorer.is_character_based()
lm_max_order = self._ext_scorer.get_max_order()
lm_dict_size = self._ext_scorer.get_dict_size()
self.logger.info("language model: "
"is_character_based = %d," % lm_char_based +
" max_order = %d," % lm_max_order +
" dict_size = %d" % lm_dict_size)
self.logger.info("end initializing scorer")
else:
self._ext_scorer = None
self.logger.info("no language model provided, "
"decoding by pure beam search without scorer.")
def _decode_batch_beam_search(self, probs_split, beam_alpha, beam_beta,
beam_size, cutoff_prob, cutoff_top_n,
vocab_list, num_processes):
"""Decode by beam search for a batch of probs matrix input.
:param probs_split: List of 2-D probability matrix, and each consists
of prob vectors for one speech utterancce.
:param probs_split: List of matrix
:param beam_alpha: Parameter associated with language model.
:type beam_alpha: float
:param beam_beta: Parameter associated with word count.
:type beam_beta: float
:param beam_size: Width for Beam search.
:type beam_size: int
:param cutoff_prob: Cutoff probability in pruning,
default 1.0, no pruning.
:type cutoff_prob: float
:param cutoff_top_n: Cutoff number in pruning, only top cutoff_top_n
characters with highest probs in vocabulary will be
used in beam search, default 40.
:type cutoff_top_n: int
:param vocab_list: List of tokens in the vocabulary, for decoding.
:type vocab_list: list
:param num_processes: Number of processes (CPU) for decoder.
:type num_processes: int
:return: List of transcription texts.
:rtype: List of str
"""
if self._ext_scorer != None:
self._ext_scorer.reset_params(beam_alpha, beam_beta)
# beam search decode
num_processes = min(num_processes, len(probs_split))
beam_search_results = ctc_beam_search_decoder_batch(
probs_split=probs_split,
vocabulary=vocab_list,
beam_size=beam_size,
num_processes=num_processes,
ext_scoring_func=self._ext_scorer,
cutoff_prob=cutoff_prob,
cutoff_top_n=cutoff_top_n)
results = [result[0][1] for result in beam_search_results]
return results
def init_decode(self, beam_alpha, beam_beta, lang_model_path, vocab_list,
decoding_method):
if decoding_method == "ctc_beam_search":
self._init_ext_scorer(beam_alpha, beam_beta, lang_model_path,
vocab_list)
def decode_probs(self, probs, logits_lens, vocab_list, decoding_method,
lang_model_path, beam_alpha, beam_beta, beam_size,
cutoff_prob, cutoff_top_n, num_processes):
""" probs: activation after softmax
logits_len: audio output lens
"""
probs_split = [probs[i, :l, :] for i, l in enumerate(logits_lens)]
if decoding_method == "ctc_greedy":
result_transcripts = self._decode_batch_greedy(
probs_split=probs_split, vocab_list=vocab_list)
elif decoding_method == "ctc_beam_search":
result_transcripts = self._decode_batch_beam_search(
probs_split=probs_split,
beam_alpha=beam_alpha,
beam_beta=beam_beta,
beam_size=beam_size,
cutoff_prob=cutoff_prob,
cutoff_top_n=cutoff_top_n,
vocab_list=vocab_list,
num_processes=num_processes)
else:
raise ValueError(f"Not support: {decoding_method}")
return result_transcripts
@paddle.no_grad()
def decode(self, audio, audio_len, vocab_list, decoding_method,
lang_model_path, beam_alpha, beam_beta, beam_size, cutoff_prob,
cutoff_top_n, num_processes):
_, probs, logits_lens = self.predict(audio, audio_len)
return self.decode_probs(probs.numpy(), logits_lens, vocab_list,
decoding_method, lang_model_path, beam_alpha,
beam_beta, beam_size, cutoff_prob,
cutoff_top_n, num_processes)
def from_pretrained(self, checkpoint_path):
"""Build a model from a pretrained model.
Parameters
----------
model: nn.Layer
Asr Model.
checkpoint_path: Path or str
The path of pretrained model checkpoint, without extension name.
Returns
-------
Model
The model build from pretrined result.
"""
checkpoint.load_parameters(self, checkpoint_path=checkpoint_path)
return
def ctc_loss(logits,
labels,
input_lengths,
label_lengths,
blank=0,
reduction='mean',
norm_by_times=True):
#logger.info("my ctc loss with norm by times")
## https://github.com/PaddlePaddle/Paddle/blob/f5ca2db2cc/paddle/fluid/operators/warpctc_op.h#L403
loss_out = paddle.fluid.layers.warpctc(logits, labels, blank, norm_by_times,
input_lengths, label_lengths)
loss_out = paddle.fluid.layers.squeeze(loss_out, [-1])
logger.info(f"warpctc loss: {loss_out}/{loss_out.shape} ")
assert reduction in ['mean', 'sum', 'none']
if reduction == 'mean':
loss_out = paddle.mean(loss_out / label_lengths)
elif reduction == 'sum':
loss_out = paddle.sum(loss_out)
logger.info(f"ctc loss: {loss_out}")
return loss_out
F.ctc_loss = ctc_loss
class DeepSpeech2Loss(nn.Layer):
def __init__(self, vocab_size):
super().__init__()
# last token id as blank id
self.loss = nn.CTCLoss(blank=vocab_size, reduction='sum')
def forward(self, logits, text, logits_len, text_len):
# warp-ctc do softmax on activations
# warp-ctc need activation with shape [T, B, V + 1]
logits = logits.transpose([1, 0, 2])
ctc_loss = self.loss(logits, text, logits_len, text_len)
return ctc_loss

@ -11,19 +11,20 @@
# 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.
"""Set up paths for DS2"""
import os.path
import sys
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
def add_path(path):
if path not in sys.path:
sys.path.insert(0, path)
logger = logging.getLogger(__name__)
__all__ = ['brelu']
this_dir = os.path.dirname(__file__)
# Add project path to PYTHONPATH
proj_path = os.path.join(this_dir, '..')
add_path(proj_path)
def brelu(x, t_min=0.0, t_max=24.0, name=None):
t_min = paddle.to_tensor(t_min)
t_max = paddle.to_tensor(t_max)
return x.maximum(t_min).minimum(t_max)

@ -0,0 +1,147 @@
# 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.
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
from deepspeech.modules.mask import sequence_mask
from deepspeech.modules.activation import brelu
logger = logging.getLogger(__name__)
__all__ = ['ConvStack']
class ConvBn(nn.Layer):
"""Convolution layer with batch normalization.
:param kernel_size: The x dimension of a filter kernel. Or input a tuple for
two image dimension.
:type kernel_size: int|tuple|list
:param num_channels_in: Number of input channels.
:type num_channels_in: int
:param num_channels_out: Number of output channels.
:type num_channels_out: int
:param stride: The x dimension of the stride. Or input a tuple for two
image dimension.
:type stride: int|tuple|list
:param padding: The x dimension of the padding. Or input a tuple for two
image dimension.
:type padding: int|tuple|list
:param act: Activation type, relu|brelu
:type act: string
:return: Batch norm layer after convolution layer.
:rtype: Variable
"""
def __init__(self, num_channels_in, num_channels_out, kernel_size, stride,
padding, act):
super().__init__()
assert len(kernel_size) == 2
assert len(stride) == 2
assert len(padding) == 2
self.kernel_size = kernel_size
self.stride = stride
self.padding = padding
self.conv = nn.Conv2D(
num_channels_in,
num_channels_out,
kernel_size=kernel_size,
stride=stride,
padding=padding,
weight_attr=None,
bias_attr=False,
data_format='NCHW')
self.bn = nn.BatchNorm2D(
num_channels_out,
weight_attr=None,
bias_attr=None,
data_format='NCHW')
self.act = F.relu if act == 'relu' else brelu
def forward(self, x, x_len):
"""
x(Tensor): audio, shape [B, C, D, T]
"""
x = self.conv(x)
x = self.bn(x)
x = self.act(x)
x_len = (x_len - self.kernel_size[1] + 2 * self.padding[1]
) // self.stride[1] + 1
# reset padding part to 0
masks = sequence_mask(x_len) #[B, T]
masks = masks.unsqueeze(1).unsqueeze(1) # [B, 1, 1, T]
x = x.multiply(masks)
return x, x_len
class ConvStack(nn.Layer):
"""Convolution group with stacked convolution layers.
:param feat_size: audio feature dim.
:type feat_size: int
:param num_stacks: Number of stacked convolution layers.
:type num_stacks: int
"""
def __init__(self, feat_size, num_stacks):
super().__init__()
self.feat_size = feat_size # D
self.num_stacks = num_stacks
self.conv_in = ConvBn(
num_channels_in=1,
num_channels_out=32,
kernel_size=(41, 11), #[D, T]
stride=(2, 3),
padding=(20, 5),
act='brelu')
out_channel = 32
self.conv_stack = nn.LayerList([
ConvBn(
num_channels_in=32,
num_channels_out=out_channel,
kernel_size=(21, 11),
stride=(2, 1),
padding=(10, 5),
act='brelu') for i in range(num_stacks - 1)
])
# conv output feat_dim
output_height = (feat_size - 1) // 2 + 1
for i in range(self.num_stacks - 1):
output_height = (output_height - 1) // 2 + 1
self.output_height = out_channel * output_height
def forward(self, x, x_len):
"""
x: shape [B, C, D, T]
x_len : shape [B]
"""
x, x_len = self.conv_in(x, x_len)
for i, conv in enumerate(self.conv_stack):
x, x_len = conv(x, x_len)
return x, x_len

@ -0,0 +1,34 @@
# 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.
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
logger = logging.getLogger(__name__)
__all__ = ['sequence_mask']
def sequence_mask(x_len, max_len=None, dtype='float32'):
max_len = max_len or x_len.max()
x_len = paddle.unsqueeze(x_len, -1)
row_vector = paddle.arange(max_len)
#mask = row_vector < x_len
mask = row_vector > x_len # a bug, broadcast 的时候出错了
mask = paddle.cast(mask, dtype)
return mask

@ -0,0 +1,309 @@
# 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.
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
from deepspeech.modules.mask import sequence_mask
from deepspeech.modules.activation import brelu
logger = logging.getLogger(__name__)
__all__ = ['RNNStack']
class RNNCell(nn.RNNCellBase):
r"""
Elman RNN (SimpleRNN) cell. Given the inputs and previous states, it
computes the outputs and updates states.
The formula used is as follows:
.. math::
h_{t} & = act(x_{t} + b_{ih} + W_{hh}h_{t-1} + b_{hh})
y_{t} & = h_{t}
where :math:`act` is for :attr:`activation`.
"""
def __init__(self,
hidden_size,
activation="tanh",
weight_ih_attr=None,
weight_hh_attr=None,
bias_ih_attr=None,
bias_hh_attr=None,
name=None):
super().__init__()
std = 1.0 / math.sqrt(hidden_size)
self.weight_hh = self.create_parameter(
(hidden_size, hidden_size),
weight_hh_attr,
default_initializer=I.Uniform(-std, std))
self.bias_ih = None
self.bias_hh = self.create_parameter(
(hidden_size, ),
bias_hh_attr,
is_bias=True,
default_initializer=I.Uniform(-std, std))
self.hidden_size = hidden_size
if activation not in ["tanh", "relu", "brelu"]:
raise ValueError(
"activation for SimpleRNNCell should be tanh or relu, "
"but get {}".format(activation))
self.activation = activation
self._activation_fn = paddle.tanh \
if activation == "tanh" \
else F.relu
if activation == 'brelu':
self._activation_fn = brelu
def forward(self, inputs, states=None):
if states is None:
states = self.get_initial_states(inputs, self.state_shape)
pre_h = states
i2h = inputs
if self.bias_ih is not None:
i2h += self.bias_ih
h2h = paddle.matmul(pre_h, self.weight_hh, transpose_y=True)
if self.bias_hh is not None:
h2h += self.bias_hh
h = self._activation_fn(i2h + h2h)
return h, h
@property
def state_shape(self):
return (self.hidden_size, )
class GRUCell(nn.RNNCellBase):
r"""
Gated Recurrent Unit (GRU) RNN cell. Given the inputs and previous states,
it computes the outputs and updates states.
The formula for GRU used is as follows:
.. math::
r_{t} & = \sigma(W_{ir}x_{t} + b_{ir} + W_{hr}h_{t-1} + b_{hr})
z_{t} & = \sigma(W_{iz}x_{t} + b_{iz} + W_{hz}h_{t-1} + b_{hz})
\widetilde{h}_{t} & = \tanh(W_{ic}x_{t} + b_{ic} + r_{t} * (W_{hc}h_{t-1} + b_{hc}))
h_{t} & = z_{t} * h_{t-1} + (1 - z_{t}) * \widetilde{h}_{t}
y_{t} & = h_{t}
where :math:`\sigma` is the sigmoid fucntion, and * is the elemetwise
multiplication operator.
"""
def __init__(self,
input_size,
hidden_size,
weight_ih_attr=None,
weight_hh_attr=None,
bias_ih_attr=None,
bias_hh_attr=None,
name=None):
super().__init__()
std = 1.0 / math.sqrt(hidden_size)
self.weight_hh = self.create_parameter(
(3 * hidden_size, hidden_size),
weight_hh_attr,
default_initializer=I.Uniform(-std, std))
self.bias_ih = None
self.bias_hh = self.create_parameter(
(3 * hidden_size, ),
bias_hh_attr,
is_bias=True,
default_initializer=I.Uniform(-std, std))
self.hidden_size = hidden_size
self.input_size = input_size
self._gate_activation = F.sigmoid
self._activation = paddle.tanh
#self._activation = F.relu
def forward(self, inputs, states=None):
if states is None:
states = self.get_initial_states(inputs, self.state_shape)
pre_hidden = states
x_gates = inputs
if self.bias_ih is not None:
x_gates = x_gates + self.bias_ih
h_gates = paddle.matmul(pre_hidden, self.weight_hh, transpose_y=True)
if self.bias_hh is not None:
h_gates = h_gates + self.bias_hh
x_r, x_z, x_c = paddle.split(x_gates, num_or_sections=3, axis=1)
h_r, h_z, h_c = paddle.split(h_gates, num_or_sections=3, axis=1)
r = self._gate_activation(x_r + h_r)
z = self._gate_activation(x_z + h_z)
c = self._activation(x_c + r * h_c) # apply reset gate after mm
h = (pre_hidden - c) * z + c
# https://www.paddlepaddle.org.cn/documentation/docs/zh/api/paddle/fluid/layers/dynamic_gru_cn.html#dynamic-gru
return h, h
@property
def state_shape(self):
r"""
The `state_shape` of GRUCell is a shape `[hidden_size]` (-1 for batch
size would be automatically inserted into shape). The shape corresponds
to the shape of :math:`h_{t-1}`.
"""
return (self.hidden_size, )
class BiRNNWithBN(nn.Layer):
"""Bidirectonal simple rnn layer with sequence-wise batch normalization.
The batch normalization is only performed on input-state weights.
:param name: Name of the layer parameters.
:type name: string
:param size: Dimension of RNN cells.
:type size: int
:param share_weights: Whether to share input-hidden weights between
forward and backward directional RNNs.
:type share_weights: bool
:return: Bidirectional simple rnn layer.
:rtype: Variable
"""
def __init__(self, i_size, h_size, share_weights):
super().__init__()
self.share_weights = share_weights
if self.share_weights:
#input-hidden weights shared between bi-directional rnn.
self.fw_fc = nn.Linear(i_size, h_size, bias_attr=False)
# batch norm is only performed on input-state projection
self.fw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.bw_fc = self.fw_fc
self.bw_bn = self.fw_bn
else:
self.fw_fc = nn.Linear(i_size, h_size, bias_attr=False)
self.fw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.bw_fc = nn.Linear(i_size, h_size, bias_attr=False)
self.bw_bn = nn.BatchNorm1D(
h_size, bias_attr=None, data_format='NLC')
self.fw_cell = RNNCell(hidden_size=h_size, activation='brelu')
self.bw_cell = RNNCell(hidden_size=h_size, activation='brelu')
self.fw_rnn = nn.RNN(
self.fw_cell, is_reverse=False, time_major=False) #[B, T, D]
self.bw_rnn = nn.RNN(
self.fw_cell, is_reverse=True, time_major=False) #[B, T, D]
def forward(self, x, x_len):
# x, shape [B, T, D]
fw_x = self.fw_bn(self.fw_fc(x))
bw_x = self.bw_bn(self.bw_fc(x))
fw_x, _ = self.fw_rnn(inputs=fw_x, sequence_length=x_len)
bw_x, _ = self.bw_rnn(inputs=bw_x, sequence_length=x_len)
x = paddle.concat([fw_x, bw_x], axis=-1)
return x, x_len
class BiGRUWithBN(nn.Layer):
"""Bidirectonal gru layer with sequence-wise batch normalization.
The batch normalization is only performed on input-state weights.
:param name: Name of the layer.
:type name: string
:param input: Input layer.
:type input: Variable
:param size: Dimension of GRU cells.
:type size: int
:param act: Activation type.
:type act: string
:return: Bidirectional GRU layer.
:rtype: Variable
"""
def __init__(self, i_size, h_size, act):
super().__init__()
hidden_size = h_size * 3
self.fw_fc = nn.Linear(i_size, hidden_size, bias_attr=False)
self.fw_bn = nn.BatchNorm1D(
hidden_size, bias_attr=None, data_format='NLC')
self.bw_fc = nn.Linear(i_size, hidden_size, bias_attr=False)
self.bw_bn = nn.BatchNorm1D(
hidden_size, bias_attr=None, data_format='NLC')
self.fw_cell = GRUCell(input_size=hidden_size, hidden_size=h_size)
self.bw_cell = GRUCell(input_size=hidden_size, hidden_size=h_size)
self.fw_rnn = nn.RNN(
self.fw_cell, is_reverse=False, time_major=False) #[B, T, D]
self.bw_rnn = nn.RNN(
self.fw_cell, is_reverse=True, time_major=False) #[B, T, D]
def forward(self, x, x_len):
# x, shape [B, T, D]
fw_x = self.fw_bn(self.fw_fc(x))
bw_x = self.bw_bn(self.bw_fc(x))
fw_x, _ = self.fw_rnn(inputs=fw_x, sequence_length=x_len)
bw_x, _ = self.bw_rnn(inputs=bw_x, sequence_length=x_len)
x = paddle.concat([fw_x, bw_x], axis=-1)
return x, x_len
class RNNStack(nn.Layer):
"""RNN group with stacked bidirectional simple RNN or GRU layers.
:param input: Input layer.
:type input: Variable
:param size: Dimension of RNN cells in each layer.
:type size: int
:param num_stacks: Number of stacked rnn layers.
:type num_stacks: int
:param use_gru: Use gru if set True. Use simple rnn if set False.
:type use_gru: bool
:param share_rnn_weights: Whether to share input-hidden weights between
forward and backward directional RNNs.
It is only available when use_gru=False.
:type share_weights: bool
:return: Output layer of the RNN group.
:rtype: Variable
"""
def __init__(self, i_size, h_size, num_stacks, use_gru, share_rnn_weights):
super().__init__()
self.rnn_stacks = nn.LayerList()
for i in range(num_stacks):
if use_gru:
#default:GRU using tanh
self.rnn_stacks.append(
BiGRUWithBN(i_size=i_size, h_size=h_size, act="relu"))
else:
self.rnn_stacks.append(
BiRNNWithBN(
i_size=i_size,
h_size=h_size,
share_weights=share_rnn_weights))
i_size = h_size * 2
def forward(self, x, x_len):
"""
x: shape [B, T, D]
x_len: shpae [B]
"""
for i, rnn in enumerate(self.rnn_stacks):
x, x_len = rnn(x, x_len)
masks = sequence_mask(x_len) #[B, T]
masks = masks.unsqueeze(-1) # [B, T, 1]
x = x.multiply(masks)
return x, x_len

@ -0,0 +1,73 @@
# 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.
import logging
from paddle.fluid.dygraph import base as imperative_base
from paddle.fluid import layers
from paddle.fluid import core
logger = logging.getLogger(__name__)
class MyClipGradByGlobalNorm(paddle.nn.ClipGradByGlobalNorm):
def __init__(self, clip_norm):
super().__init__(clip_norm)
@imperative_base.no_grad
def _dygraph_clip(self, params_grads):
params_and_grads = []
sum_square_list = []
for p, g in params_grads:
if g is None:
continue
if getattr(p, 'need_clip', True) is False:
continue
merge_grad = g
if g.type == core.VarDesc.VarType.SELECTED_ROWS:
merge_grad = layers.merge_selected_rows(g)
merge_grad = layers.get_tensor_from_selected_rows(merge_grad)
square = layers.square(merge_grad)
sum_square = layers.reduce_sum(square)
logger.info(
f"Grad Before Clip: {p.name}: {float(layers.sqrt(layers.reduce_sum(layers.square(merge_grad))) ) }"
)
sum_square_list.append(sum_square)
# all parameters have been filterd out
if len(sum_square_list) == 0:
return params_grads
global_norm_var = layers.concat(sum_square_list)
global_norm_var = layers.reduce_sum(global_norm_var)
global_norm_var = layers.sqrt(global_norm_var)
logger.info(f"Grad Global Norm: {float(global_norm_var)}!!!!")
max_global_norm = layers.fill_constant(
shape=[1], dtype=global_norm_var.dtype, value=self.clip_norm)
clip_var = layers.elementwise_div(
x=max_global_norm,
y=layers.elementwise_max(x=global_norm_var, y=max_global_norm))
for p, g in params_grads:
if g is None:
continue
if getattr(p, 'need_clip', True) is False:
params_and_grads.append((p, g))
continue
new_grad = layers.elementwise_mul(x=g, y=clip_var)
logger.info(
f"Grad After Clip: {p.name}: {float(layers.sqrt(layers.reduce_sum(layers.square(merge_grad))) ) }"
)
params_and_grads.append((p, new_grad))
return params_and_grads

@ -0,0 +1,65 @@
# 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.
import logging
import paddle
from paddle import nn
from paddle.nn import functional as F
from paddle.nn import initializer as I
logger = logging.getLogger(__name__)
__all__ = ['CTCLoss']
def ctc_loss(logits,
labels,
input_lengths,
label_lengths,
blank=0,
reduction='mean',
norm_by_times=True):
#logger.info("my ctc loss with norm by times")
## https://github.com/PaddlePaddle/Paddle/blob/f5ca2db2cc/paddle/fluid/operators/warpctc_op.h#L403
loss_out = paddle.fluid.layers.warpctc(logits, labels, blank, norm_by_times,
input_lengths, label_lengths)
loss_out = paddle.fluid.layers.squeeze(loss_out, [-1])
logger.info(f"warpctc loss: {loss_out}/{loss_out.shape} ")
assert reduction in ['mean', 'sum', 'none']
if reduction == 'mean':
loss_out = paddle.mean(loss_out / label_lengths)
elif reduction == 'sum':
loss_out = paddle.sum(loss_out)
logger.info(f"ctc loss: {loss_out}")
return loss_out
F.ctc_loss = ctc_loss
class CTCLoss(nn.Layer):
def __init__(self, blank_id):
super().__init__()
# last token id as blank id
self.loss = nn.CTCLoss(blank=blank_id, reduction='sum')
def forward(self, logits, text, logits_len, text_len):
# warp-ctc do softmax on activations
# warp-ctc need activation with shape [T, B, V + 1]
logits = logits.transpose([1, 0, 2])
ctc_loss = self.loss(logits, text, logits_len, text_len)
return ctc_loss

@ -15,6 +15,8 @@
import distutils.util
__all__ = ['print_arguments', 'add_arguments', 'print_grads', 'print_params']
def print_arguments(args):
"""Print argparse's arguments.
@ -55,3 +57,21 @@ def add_arguments(argname, type, default, help, argparser, **kwargs):
type=type,
help=help + ' Default: %(default)s.',
**kwargs)
def print_grads(model, logger=None):
for n, p in model.named_parameters():
msg = f"param grad: {n}: shape: {p.shape} grad: {p.grad}"
if logger:
logger.info(msg)
def print_params(model, logger=None):
total = 0.0
for n, p in model.named_parameters():
msg = f"param: {n}: shape: {p.shape} stop_grad: {p.stop_gradient}"
total += np.prod(p.shape)
if logger:
logger.info(msg)
if logger:
logger.info(f"Total parameters: {total}!")

Loading…
Cancel
Save