# Copyright (c) 2020, NVIDIA CORPORATION. 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 io
import os
from typing import Callable, Dict, List, Optional, Union
import braceexpand
import torch
import webdataset as wd
from torch.nn import functional as F
from nemo.collections.asr.data import vocabs
from nemo.collections.asr.parts import collections, parsers
from nemo.collections.asr.parts.features import WaveformFeaturizer
from nemo.core.classes import Dataset, IterableDataset
from nemo.core.neural_types import *
from nemo.utils import logging
__all__ = [
'AudioToCharDataset',
'AudioToCharWithDursDataset',
'AudioToBPEDataset',
'TarredAudioToCharDataset',
'TarredAudioToBPEDataset',
]
def _speech_collate_fn(batch, pad_id):
"""collate batch of audio sig, audio len, tokens, tokens len
Args:
batch (Optional[FloatTensor], Optional[LongTensor], LongTensor,
LongTensor): A tuple of tuples of signal, signal lengths,
encoded tokens, and encoded tokens length. This collate func
assumes the signals are 1d torch tensors (i.e. mono audio).
"""
_, audio_lengths, _, tokens_lengths = zip(*batch)
max_audio_len = 0
has_audio = audio_lengths[0] is not None
if has_audio:
max_audio_len = max(audio_lengths).item()
max_tokens_len = max(tokens_lengths).item()
audio_signal, tokens = [], []
for sig, sig_len, tokens_i, tokens_i_len in batch:
if has_audio:
sig_len = sig_len.item()
if sig_len < max_audio_len:
pad = (0, max_audio_len - sig_len)
sig = torch.nn.functional.pad(sig, pad)
audio_signal.append(sig)
tokens_i_len = tokens_i_len.item()
if tokens_i_len < max_tokens_len:
pad = (0, max_tokens_len - tokens_i_len)
tokens_i = torch.nn.functional.pad(tokens_i, pad, value=pad_id)
tokens.append(tokens_i)
if has_audio:
audio_signal = torch.stack(audio_signal)
audio_lengths = torch.stack(audio_lengths)
else:
audio_signal, audio_lengths = None, None
tokens = torch.stack(tokens)
tokens_lengths = torch.stack(tokens_lengths)
return audio_signal, audio_lengths, tokens, tokens_lengths
class _AudioTextDataset(Dataset):
"""
Dataset that loads tensors via a json file containing paths to audio files, transcripts, and durations (in seconds).
Each new line is a different sample. Example below:
{"audio_filepath": "/path/to/audio.wav", "text_filepath": "/path/to/audio.txt", "duration": 23.147}
...
{"audio_filepath": "/path/to/audio.wav", "text": "the transcription", "offset": 301.75, "duration": 0.82, "utt":
"utterance_id", "ctm_utt": "en_4156", "side": "A"}
Args:
manifest_filepath: Path to manifest json as described above. Can be comma-separated paths.
labels: String containing all the possible characters to map to
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor object used to augment loaded
audio
max_duration: If audio exceeds this length, do not include in dataset
min_duration: If audio is less than this length, do not include in dataset
max_utts: Limit number of utterances
blank_index: blank character index, default = -1
unk_index: unk_character index, default = -1
normalize: whether to normalize transcript text (default): True
bos_id: Id of beginning of sequence symbol to append if not None
eos_id: Id of end of sequence symbol to append if not None
load_audio: Boolean flag indicate whether do or not load audio
add_misc: True if add additional info dict.
"""
@property
def output_types(self) -> Optional[Dict[str, NeuralType]]:
"""Returns definitions of module output ports.
"""
return {
'audio_signal': NeuralType(
('B', 'T'),
AudioSignal(freq=self._sample_rate) # TODO: self._sample_rate is not defined anywhere
if self is not None and hasattr(self, '_sample_rate')
else AudioSignal(),
),
'a_sig_length': NeuralType(tuple('B'), LengthsType()),
'transcripts': NeuralType(('B', 'T'), LabelsType()),
'transcript_length': NeuralType(tuple('B'), LengthsType()),
}
def __init__(
self,
manifest_filepath: str,
parser: Union[str, Callable],
sample_rate: int,
int_values: bool = False,
augmentor: 'nemo.collections.asr.parts.perturb.AudioAugmentor' = None,
max_duration: Optional[int] = None,
min_duration: Optional[int] = None,
max_utts: int = 0,
trim: bool = False,
bos_id: Optional[int] = None,
eos_id: Optional[int] = None,
pad_id: int = 0,
load_audio: bool = True,
add_misc: bool = False,
):
self.parser = parser
self.collection = collections.ASRAudioText(
manifests_files=manifest_filepath.split(','),
parser=parser,
min_duration=min_duration,
max_duration=max_duration,
max_number=max_utts,
)
self.featurizer = WaveformFeaturizer(sample_rate=sample_rate, int_values=int_values, augmentor=augmentor)
self.trim = trim
self.eos_id = eos_id
self.bos_id = bos_id
self.pad_id = pad_id
self.load_audio = load_audio
self._add_misc = add_misc
def __getitem__(self, index):
sample = self.collection[index]
if self.load_audio:
offset = sample.offset
if offset is None:
offset = 0
features = self.featurizer.process(
sample.audio_file, offset=offset, duration=sample.duration, trim=self.trim, orig_sr=sample.orig_sr
)
f, fl = features, torch.tensor(features.shape[0]).long()
else:
f, fl = None, None
t, tl = sample.text_tokens, len(sample.text_tokens)
if self.bos_id is not None:
t = [self.bos_id] + t
tl += 1
if self.eos_id is not None:
t = t + [self.eos_id]
tl += 1
output = f, fl, torch.tensor(t).long(), torch.tensor(tl).long()
if self._add_misc:
misc = dict()
misc['id'] = sample.id
misc['text_raw'] = sample.text_raw
misc['speaker'] = sample.speaker
output = (output, misc)
return output
def __len__(self):
return len(self.collection)
def _collate_fn(self, batch):
return _speech_collate_fn(batch, pad_id=self.pad_id)
[docs]class AudioToCharDataset(_AudioTextDataset):
"""
Dataset that loads tensors via a json file containing paths to audio
files, transcripts, and durations (in seconds). Each new line is a
different sample. Example below:
{"audio_filepath": "/path/to/audio.wav", "text_filepath":
"/path/to/audio.txt", "duration": 23.147}
...
{"audio_filepath": "/path/to/audio.wav", "text": "the
transcription", "offset": 301.75, "duration": 0.82, "utt":
"utterance_id", "ctm_utt": "en_4156", "side": "A"}
Args:
manifest_filepath: Path to manifest json as described above. Can
be comma-separated paths.
labels: String containing all the possible characters to map to
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor
object used to augment loaded audio
max_duration: If audio exceeds this length, do not include in dataset
min_duration: If audio is less than this length, do not include
in dataset
max_utts: Limit number of utterances
blank_index: blank character index, default = -1
unk_index: unk_character index, default = -1
normalize: whether to normalize transcript text (default): True
bos_id: Id of beginning of sequence symbol to append if not None
eos_id: Id of end of sequence symbol to append if not None
load_audio: Boolean flag indicate whether do or not load audio
add_misc: True if add additional info dict.
"""
@property
def output_types(self) -> Optional[Dict[str, NeuralType]]:
"""Returns definitions of module output ports.
"""
return {
'audio_signal': NeuralType(
('B', 'T'),
AudioSignal(freq=self._sample_rate)
if self is not None and hasattr(self, '_sample_rate')
else AudioSignal(),
),
'a_sig_length': NeuralType(tuple('B'), LengthsType()),
'transcripts': NeuralType(('B', 'T'), LabelsType()),
'transcript_length': NeuralType(tuple('B'), LengthsType()),
}
def __init__(
self,
manifest_filepath: str,
labels: Union[str, List[str]],
sample_rate: int,
int_values: bool = False,
augmentor: 'nemo.collections.asr.parts.perturb.AudioAugmentor' = None,
max_duration: Optional[float] = None,
min_duration: Optional[float] = None,
max_utts: int = 0,
blank_index: int = -1,
unk_index: int = -1,
normalize: bool = True,
trim: bool = False,
bos_id: Optional[int] = None,
eos_id: Optional[int] = None,
pad_id: int = 0,
load_audio: bool = True,
parser: Union[str, Callable] = 'en',
add_misc: bool = False,
):
self.labels = labels
parser = parsers.make_parser(
labels=labels, name=parser, unk_id=unk_index, blank_id=blank_index, do_normalize=normalize
)
super().__init__(
manifest_filepath=manifest_filepath,
parser=parser,
sample_rate=sample_rate,
int_values=int_values,
augmentor=augmentor,
max_duration=max_duration,
min_duration=min_duration,
max_utts=max_utts,
trim=trim,
bos_id=bos_id,
eos_id=eos_id,
pad_id=pad_id,
load_audio=load_audio,
add_misc=add_misc,
)
[docs]class AudioToCharWithDursDataset(AudioToCharDataset):
"""
Dataset that loads tensors via a json file containing paths to audio
files, transcripts, and durations (in seconds). Each new line is a
different sample. Example below:
{"audio_filepath": "/path/to/audio.wav", "text_filepath":
"/path/to/audio.txt", "duration": 23.147}
...
{"audio_filepath": "/path/to/audio.wav", "text": "the
transcription", "offset": 301.75, "duration": 0.82, "utt":
"utterance_id", "ctm_utt": "en_4156", "side": "A"}
Additionally, user provides path to precomputed durations, which is a pickled python dict with 'tags' and 'durs'
keys, both of which are list of examples values. Tag is a unique example identifier, which is a wav filename
without suffix. Durations are an additional tuple of two tensors: graphemes durations and blanks durations.
Example below:
{'tags': ['LJ050-0234', 'LJ019-0373'],
'durs': [(graphemes_durs0, blanks_durs0), (graphemes_durs1, blanks_durs1)]}
Args:
**kwargs: Passed to AudioToCharDataset constructor.
durs_path (str): String path to pickled list of '[(tag, durs)]' durations location.
rep (bool): True if repeat text graphemes according to durs.
vocab: Vocabulary config (parser + set of graphemes to use). Constructor propagates these to
`self.make_vocab` function call to build a complete vocabulary.
"""
@property
def output_types(self) -> Optional[Dict[str, NeuralType]]:
"""Returns definitions of module output ports."""
return {
'audio': NeuralType(
('B', 'T'),
AudioSignal(freq=self._sample_rate)
if self is not None and hasattr(self, '_sample_rate')
else AudioSignal(),
),
'audio_len': NeuralType(('B',), LengthsType()),
'text': NeuralType(('B', 'T'), LabelsType()),
'text_len': NeuralType(('B',), LengthsType()),
'durs': NeuralType(('B', 'T'), LengthsType()),
}
[docs] @staticmethod
def make_vocab(notation='chars', punct=True, spaces=False, stresses=False):
"""Constructs vocabulary from given parameters.
Args:
notation (str): Either 'chars' or 'phonemes' as general notation.
punct (bool): True if reserve grapheme for basic punctuation.
spaces (bool): True if prepend spaces to every punctuation symbol.
stresses (bool): True if use phonemes codes with stresses (0-2).
Returns:
(vocabs.Base) Vocabulary
"""
if notation == 'chars':
vocab = vocabs.Chars(punct=punct, spaces=spaces)
elif notation == 'phonemes':
vocab = vocabs.Phonemes(punct=punct, stresses=stresses, spaces=spaces)
else:
raise ValueError("Unsupported vocab type.")
return vocab
def __init__(self, **kwargs):
durs_path = kwargs.pop('durs_path')
rep = kwargs.pop('rep', False)
self.vocab = self.make_vocab(**kwargs.pop('vocab', {}))
kwargs.setdefault('labels', [])
super().__init__(**kwargs)
pth = torch.load(durs_path)
tag2d = dict(zip(pth['tags'], pth['durs']))
durs = []
for i, e in enumerate(self.collection):
tag = os.path.splitext(os.path.basename(e.audio_file))[0]
durs.append(tag2d[tag])
self.durs = durs
self.rep = rep
def __getitem__(self, item):
sample = self.collection[item]
audio, audio_len, _, _ = super().__getitem__(item) # noqa
text = self.vocab.encode(sample.text_raw)
text, text_len = torch.tensor(text).long(), torch.tensor(len(text)).long()
blanks_durs, graphemes_durs = self.durs[item]
return (
audio,
audio_len,
text,
text_len,
blanks_durs,
graphemes_durs,
)
@staticmethod
def _merge(tensors, dim=0, value=0, dtype=None):
"""Merges list of tensors into one."""
tensors = [tensor if isinstance(tensor, torch.Tensor) else torch.tensor(tensor) for tensor in tensors]
dim = dim if dim != -1 else len(tensors[0].shape) - 1
dtype = tensors[0].dtype if dtype is None else dtype
max_len = max(tensor.shape[dim] for tensor in tensors)
new_tensors = []
for tensor in tensors:
pad = (2 * len(tensor.shape)) * [0]
pad[-2 * dim - 1] = max_len - tensor.shape[dim]
new_tensors.append(F.pad(tensor, pad=pad, value=value))
return torch.stack(new_tensors).to(dtype=dtype)
@staticmethod
def _interleave(x, y):
"""Interleave two tensors."""
xy = torch.stack([x[:-1], y], dim=1).view(-1)
xy = F.pad(xy, pad=[0, 1], value=x[-1])
return xy
def _collate_fn(self, batch):
batch = list(zip(*batch))
asr_batch = _speech_collate_fn(list(zip(*batch[:4])), pad_id=self.vocab.pad)
audio, audio_len, text, text_len = asr_batch
text = [
self._interleave(
x=torch.empty(len(t) + 1, dtype=torch.long, device=t.device,).fill_(self.vocab.blank), y=t,
)
for t in text
]
text = self._merge(text, value=self.vocab.pad, dtype=torch.long)
text_len = text_len * 2 + 1
blanks_durs, graphemes_durs = batch[4:]
durs = [self._interleave(b, c) for b, c in zip(blanks_durs, graphemes_durs)]
durs = self._merge(durs, dtype=torch.long).to(text.device)
if self.rep:
text = self._merge(
tensors=[torch.repeat_interleave(text1, durs1) for text1, durs1 in zip(text, durs)], dtype=torch.long,
)
text_len = durs.sum(-1)
return (
audio,
audio_len,
text,
text_len,
durs,
)
[docs]class AudioToBPEDataset(_AudioTextDataset):
"""
Dataset that loads tensors via a json file containing paths to audio
files, transcripts, and durations (in seconds). Each new line is a
different sample. Example below:
{"audio_filepath": "/path/to/audio.wav", "text_filepath":
"/path/to/audio.txt", "duration": 23.147}
...
{"audio_filepath": "/path/to/audio.wav", "text": "the
transcription", "offset": 301.75, "duration": 0.82, "utt":
"utterance_id", "ctm_utt": "en_4156", "side": "A"}
In practice, the dataset and manifest used for character encoding and byte pair encoding
are exactly the same. The only difference lies in how the dataset tokenizes the text in
the manifest.
Args:
manifest_filepath: Path to manifest json as described above. Can
be comma-separated paths.
tokenizer: A subclass of the Tokenizer wrapper found in the common collection,
nemo.collections.common.tokenizers.TokenizerSpec. ASR Models support a subset of
all available tokenizers.
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor
object used to augment loaded audio
max_duration: If audio exceeds this length, do not include in dataset
min_duration: If audio is less than this length, do not include
in dataset
max_utts: Limit number of utterances
trim: Whether to trim silence segments
load_audio: Boolean flag indicate whether do or not load audio
add_misc: True if add additional info dict.
use_start_end_token: Boolean which dictates whether to add [BOS] and [EOS]
tokens to beginning and ending of speech respectively.
"""
@property
def output_types(self) -> Optional[Dict[str, NeuralType]]:
"""Returns definitions of module output ports.
"""
return {
'audio_signal': NeuralType(
('B', 'T'),
AudioSignal(freq=self._sample_rate)
if self is not None and hasattr(self, '_sample_rate')
else AudioSignal(),
),
'a_sig_length': NeuralType(tuple('B'), LengthsType()),
'transcripts': NeuralType(('B', 'T'), LabelsType()),
'transcript_length': NeuralType(tuple('B'), LengthsType()),
}
def __init__(
self,
manifest_filepath: str,
tokenizer: 'nemo.collections.common.tokenizers.TokenizerSpec',
sample_rate: int,
int_values: bool = False,
augmentor: 'nemo.collections.asr.parts.perturb.AudioAugmentor' = None,
max_duration: Optional[int] = None,
min_duration: Optional[int] = None,
max_utts: int = 0,
trim: bool = False,
load_audio: bool = True,
add_misc: bool = False,
use_start_end_token: bool = True,
):
if use_start_end_token and hasattr(tokenizer, 'bos_token'):
bos_id = tokenizer.bos_id
else:
bos_id = None
if use_start_end_token and hasattr(tokenizer, 'eos_token'):
eos_id = tokenizer.eos_id
else:
eos_id = None
if hasattr(tokenizer, 'pad_token'):
pad_id = tokenizer.pad_id
else:
pad_id = 0
class TokenizerWrapper:
def __init__(self, tokenizer):
self._tokenizer = tokenizer
def __call__(self, text):
t = self._tokenizer.text_to_ids(text)
return t
super().__init__(
manifest_filepath=manifest_filepath,
parser=TokenizerWrapper(tokenizer),
sample_rate=sample_rate,
int_values=int_values,
augmentor=augmentor,
max_duration=max_duration,
min_duration=min_duration,
max_utts=max_utts,
bos_id=bos_id,
eos_id=eos_id,
pad_id=pad_id,
trim=trim,
load_audio=load_audio,
add_misc=add_misc,
)
class _TarredAudioToTextDataset(IterableDataset):
"""
A similar Dataset to the AudioToCharDataset/AudioToBPEDataset, but which loads tarred audio files.
Accepts a single comma-separated JSON manifest file (in the same style as for the AudioToCharDataset/AudioToBPEDataset),
as well as the path(s) to the tarball(s) containing the wav files. Each line of the manifest should
contain the information for one audio file, including at least the transcript and name of the audio
file within the tarball.
Valid formats for the audio_tar_filepaths argument include:
(1) a single string that can be brace-expanded, e.g. 'path/to/audio.tar' or 'path/to/audio_{1..100}.tar.gz', or
(2) a list of file paths that will not be brace-expanded, e.g. ['audio_1.tar', 'audio_2.tar', ...].
Note: For brace expansion in (1), there may be cases where `{x..y}` syntax cannot be used due to shell interference.
This occurs most commonly inside SLURM scripts. Therefore we provide a few equivalent replacements.
Supported opening braces - { <=> (, [, < and the special tag _OP_.
Supported closing braces - } <=> ), ], > and the special tag _CL_.
For SLURM based tasks, we suggest the use of the special tags for ease of use.
See the WebDataset documentation for more information about accepted data and input formats.
If using multiple workers the number of shards should be divisible by world_size to ensure an
even split among workers. If it is not divisible, logging will give a warning but training will proceed.
In addition, if using mutiprocessing, each shard MUST HAVE THE SAME NUMBER OF ENTRIES after filtering
is applied. We currently do not check for this, but your program may hang if the shards are uneven!
Notice that a few arguments are different from the AudioToCharDataset; for example, shuffle (bool) has been
replaced by shuffle_n (int).
Additionally, please note that the len() of this DataLayer is assumed to be the length of the manifest
after filtering. An incorrect manifest length may lead to some DataLoader issues down the line.
Args:
audio_tar_filepaths: Either a list of audio tarball filepaths, or a
string (can be brace-expandable).
manifest_filepath (str): Path to the manifest.
parser (callable): A callable which is used to pre-process the text output.
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor
object used to augment loaded audio
shuffle_n (int): How many samples to look ahead and load to be shuffled.
See WebDataset documentation for more details.
Defaults to 0.
min_duration (float): Dataset parameter.
All training files which have a duration less than min_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to 0.1.
max_duration (float): Dataset parameter.
All training files which have a duration more than max_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to None.
max_utts (int): Limit number of utterances. 0 means no maximum.
blank_index (int): Blank character index, defaults to -1.
unk_index (int): Unknown character index, defaults to -1.
normalize (bool): Dataset parameter.
Whether to use automatic text cleaning.
It is highly recommended to manually clean text for best results.
Defaults to True.
trim (bool): Whether to use trim silence from beginning and end
of audio signal using librosa.effects.trim().
Defaults to False.
bos_id (id): Dataset parameter.
Beginning of string symbol id used for seq2seq models.
Defaults to None.
eos_id (id): Dataset parameter.
End of string symbol id used for seq2seq models.
Defaults to None.
pad_id (id): Token used to pad when collating samples in batches.
If this is None, pads using 0s.
Defaults to None.
shard_strategy (str): Tarred dataset shard distribution strategy chosen as a str value during ddp.
- `scatter`: The default shard strategy applied by WebDataset, where each node gets
a unique set of shards, which are permanently pre-allocated and never changed at runtime.
- `replicate`: Optional shard strategy, where each node gets all of the set of shards
available in the tarred dataset, which are permanently pre-allocated and never changed at runtime.
The benefit of replication is that it allows each node to sample data points from the entire
dataset independently of other nodes, and reduces dependence on value of `shuffle_n`.
Note: Replicated strategy allows every node to sample the entire set of available tarfiles,
and therefore more than one node may sample the same tarfile, and even sample the same
data points! As such, there is no assured guarantee that all samples in the dataset will be
sampled at least once during 1 epoch.
global_rank (int): Worker rank, used for partitioning shards. Defaults to 0.
world_size (int): Total number of processes, used for partitioning shards. Defaults to 0.
"""
def __init__(
self,
audio_tar_filepaths: Union[str, List[str]],
manifest_filepath: str,
parser: Callable,
sample_rate: int,
int_values: bool = False,
augmentor: Optional['nemo.collections.asr.parts.perturb.AudioAugmentor'] = None,
shuffle_n: int = 0,
min_duration: Optional[float] = None,
max_duration: Optional[float] = None,
max_utts: int = 0,
trim: bool = False,
bos_id: Optional[int] = None,
eos_id: Optional[int] = None,
add_misc: bool = False,
pad_id: int = 0,
shard_strategy: str = "scatter",
global_rank: int = 0,
world_size: int = 0,
):
self.collection = collections.ASRAudioText(
manifests_files=manifest_filepath.split(','),
parser=parser,
min_duration=min_duration,
max_duration=max_duration,
max_number=max_utts,
index_by_file_id=True, # Must set this so the manifest lines can be indexed by file ID
)
self.featurizer = WaveformFeaturizer(sample_rate=sample_rate, int_values=int_values, augmentor=augmentor)
self.trim = trim
self.eos_id = eos_id
self.bos_id = bos_id
self.pad_id = pad_id
self._add_misc = add_misc
valid_shard_strategies = ['scatter', 'replicate']
if shard_strategy not in valid_shard_strategies:
raise ValueError(f"`shard_strategy` must be one of {valid_shard_strategies}")
if isinstance(audio_tar_filepaths, str):
# Replace '(' and '[' with '{'
brace_keys_open = ['(', '[', '<', '_OP_']
for bkey in brace_keys_open:
if bkey in audio_tar_filepaths:
audio_tar_filepaths = audio_tar_filepaths.replace(bkey, "{")
# Replace ')' and ']' with '}'
brace_keys_close = [')', ']', '>', '_CL_']
for bkey in brace_keys_close:
if bkey in audio_tar_filepaths:
audio_tar_filepaths = audio_tar_filepaths.replace(bkey, "}")
# Check for distributed and partition shards accordingly
if world_size > 1:
if isinstance(audio_tar_filepaths, str):
# Brace expand
audio_tar_filepaths = list(braceexpand.braceexpand(audio_tar_filepaths))
if shard_strategy == 'scatter':
logging.info("All tarred dataset shards will be scattered evenly across all nodes.")
if len(audio_tar_filepaths) % world_size != 0:
logging.warning(
f"Number of shards in tarred dataset ({len(audio_tar_filepaths)}) is not divisible "
f"by number of distributed workers ({world_size})."
)
begin_idx = (len(audio_tar_filepaths) // world_size) * global_rank
end_idx = begin_idx + (len(audio_tar_filepaths) // world_size)
audio_tar_filepaths = audio_tar_filepaths[begin_idx:end_idx]
logging.info(
"Partitioning tarred dataset: process (%d) taking shards [%d, %d)", global_rank, begin_idx, end_idx
)
elif shard_strategy == 'replicate':
logging.info("All tarred dataset shards will be replicated across all nodes.")
else:
raise ValueError(f"Invalid shard strategy ! Allowed values are : {valid_shard_strategies}")
# Put together WebDataset
self._dataset = wd.WebDataset(audio_tar_filepaths)
if shuffle_n > 0:
self._dataset = self._dataset.shuffle(shuffle_n)
else:
logging.info("WebDataset will not shuffle files within the tar files.")
self._dataset = (
self._dataset.rename(audio='wav', key='__key__')
.to_tuple('audio', 'key')
.pipe(self._filter)
.map(f=self._build_sample)
)
def _filter(self, iterator):
"""This function is used to remove samples that have been filtered out by ASRAudioText already.
Otherwise, we would get a KeyError as _build_sample attempts to find the manifest entry for a sample
that was filtered out (e.g. for duration).
Note that if using multi-GPU training, filtering may lead to an imbalance in samples in each shard,
which may make your code hang as one process will finish before the other.
"""
class TarredAudioFilter:
def __init__(self, collection):
self.iterator = iterator
self.collection = collection
def __iter__(self):
return self
def __next__(self):
while True:
audio_bytes, audio_filename = next(self.iterator)
file_id, _ = os.path.splitext(os.path.basename(audio_filename))
if file_id in self.collection.mapping:
return audio_bytes, audio_filename
return TarredAudioFilter(self.collection)
def _collate_fn(self, batch):
return _speech_collate_fn(batch, self.pad_id)
def _build_sample(self, tup):
"""Builds the training sample by combining the data from the WebDataset with the manifest info.
"""
audio_bytes, audio_filename = tup
# Grab manifest entry from self.collection
file_id, _ = os.path.splitext(os.path.basename(audio_filename))
manifest_idx = self.collection.mapping[file_id]
manifest_entry = self.collection[manifest_idx]
offset = manifest_entry.offset
if offset is None:
offset = 0
# Convert audio bytes to IO stream for processing (for SoundFile to read)
audio_filestream = io.BytesIO(audio_bytes)
features = self.featurizer.process(
audio_filestream,
offset=offset,
duration=manifest_entry.duration,
trim=self.trim,
orig_sr=manifest_entry.orig_sr,
)
audio_filestream.close()
# Audio features
f, fl = features, torch.tensor(features.shape[0]).long()
# Text features
t, tl = manifest_entry.text_tokens, len(manifest_entry.text_tokens)
if self.bos_id is not None:
t = [self.bos_id] + t
tl += 1
if self.eos_id is not None:
t = t + [self.eos_id]
tl += 1
return f, fl, torch.tensor(t).long(), torch.tensor(tl).long()
def __iter__(self):
return self._dataset.__iter__()
def __len__(self):
return len(self.collection)
[docs]class TarredAudioToCharDataset(_TarredAudioToTextDataset):
"""
A similar Dataset to the AudioToCharDataset, but which loads tarred audio files.
Accepts a single comma-separated JSON manifest file (in the same style as for the AudioToCharDataset),
as well as the path(s) to the tarball(s) containing the wav files. Each line of the manifest should
contain the information for one audio file, including at least the transcript and name of the audio
file within the tarball.
Valid formats for the audio_tar_filepaths argument include:
(1) a single string that can be brace-expanded, e.g. 'path/to/audio.tar' or 'path/to/audio_{1..100}.tar.gz', or
(2) a list of file paths that will not be brace-expanded, e.g. ['audio_1.tar', 'audio_2.tar', ...].
See the WebDataset documentation for more information about accepted data and input formats.
If using multiple workers the number of shards should be divisible by world_size to ensure an
even split among workers. If it is not divisible, logging will give a warning but training will proceed.
In addition, if using mutiprocessing, each shard MUST HAVE THE SAME NUMBER OF ENTRIES after filtering
is applied. We currently do not check for this, but your program may hang if the shards are uneven!
Notice that a few arguments are different from the AudioToCharDataset; for example, shuffle (bool) has been
replaced by shuffle_n (int).
Additionally, please note that the len() of this DataLayer is assumed to be the length of the manifest
after filtering. An incorrect manifest length may lead to some DataLoader issues down the line.
Args:
audio_tar_filepaths: Either a list of audio tarball filepaths, or a
string (can be brace-expandable).
manifest_filepath (str): Path to the manifest.
labels (list): List of characters that can be output by the ASR model.
For Jasper, this is the 28 character set {a-z '}. The CTC blank
symbol is automatically added later for models using ctc.
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor
object used to augment loaded audio
shuffle_n (int): How many samples to look ahead and load to be shuffled.
See WebDataset documentation for more details.
Defaults to 0.
min_duration (float): Dataset parameter.
All training files which have a duration less than min_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to 0.1.
max_duration (float): Dataset parameter.
All training files which have a duration more than max_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to None.
max_utts (int): Limit number of utterances. 0 means no maximum.
blank_index (int): Blank character index, defaults to -1.
unk_index (int): Unknown character index, defaults to -1.
normalize (bool): Dataset parameter.
Whether to use automatic text cleaning.
It is highly recommended to manually clean text for best results.
Defaults to True.
trim (bool): Whether to use trim silence from beginning and end
of audio signal using librosa.effects.trim().
Defaults to False.
bos_id (id): Dataset parameter.
Beginning of string symbol id used for seq2seq models.
Defaults to None.
eos_id (id): Dataset parameter.
End of string symbol id used for seq2seq models.
Defaults to None.
pad_id (id): Token used to pad when collating samples in batches.
If this is None, pads using 0s.
Defaults to None.
shard_strategy (str): Tarred dataset shard distribution strategy chosen as a str value during ddp.
- `scatter`: The default shard strategy applied by WebDataset, where each node gets
a unique set of shards, which are permanently pre-allocated and never changed at runtime.
- `replicate`: Optional shard strategy, where each node gets all of the set of shards
available in the tarred dataset, which are permanently pre-allocated and never changed at runtime.
The benefit of replication is that it allows each node to sample data points from the entire
dataset independently of other nodes, and reduces dependence on value of `shuffle_n`.
Note: Replicated strategy allows every node to sample the entire set of available tarfiles,
and therefore more than one node may sample the same tarfile, and even sample the same
data points! As such, there is no assured guarantee that all samples in the dataset will be
sampled at least once during 1 epoch.
global_rank (int): Worker rank, used for partitioning shards. Defaults to 0.
world_size (int): Total number of processes, used for partitioning shards. Defaults to 0.
"""
def __init__(
self,
audio_tar_filepaths: Union[str, List[str]],
manifest_filepath: str,
labels: List[str],
sample_rate: int,
int_values: bool = False,
augmentor: Optional['nemo.collections.asr.parts.perturb.AudioAugmentor'] = None,
shuffle_n: int = 0,
min_duration: Optional[float] = None,
max_duration: Optional[float] = None,
max_utts: int = 0,
blank_index: int = -1,
unk_index: int = -1,
normalize: bool = True,
trim: bool = False,
bos_id: Optional[int] = None,
eos_id: Optional[int] = None,
parser: Optional[str] = 'en',
add_misc: bool = False,
pad_id: int = 0,
shard_strategy: str = "scatter",
global_rank: int = 0,
world_size: int = 0,
):
self.labels = labels
parser = parsers.make_parser(
labels=labels, name=parser, unk_id=unk_index, blank_id=blank_index, do_normalize=normalize
)
super().__init__(
audio_tar_filepaths=audio_tar_filepaths,
manifest_filepath=manifest_filepath,
parser=parser,
sample_rate=sample_rate,
int_values=int_values,
augmentor=augmentor,
shuffle_n=shuffle_n,
min_duration=min_duration,
max_duration=max_duration,
max_utts=max_utts,
trim=trim,
bos_id=bos_id,
eos_id=eos_id,
add_misc=add_misc,
pad_id=pad_id,
shard_strategy=shard_strategy,
global_rank=global_rank,
world_size=world_size,
)
[docs]class TarredAudioToBPEDataset(_TarredAudioToTextDataset):
"""
A similar Dataset to the AudioToBPEDataset, but which loads tarred audio files.
Accepts a single comma-separated JSON manifest file (in the same style as for the AudioToBPEDataset),
as well as the path(s) to the tarball(s) containing the wav files. Each line of the manifest should
contain the information for one audio file, including at least the transcript and name of the audio
file within the tarball.
Valid formats for the audio_tar_filepaths argument include:
(1) a single string that can be brace-expanded, e.g. 'path/to/audio.tar' or 'path/to/audio_{1..100}.tar.gz', or
(2) a list of file paths that will not be brace-expanded, e.g. ['audio_1.tar', 'audio_2.tar', ...].
See the WebDataset documentation for more information about accepted data and input formats.
If using multiple workers the number of shards should be divisible by world_size to ensure an
even split among workers. If it is not divisible, logging will give a warning but training will proceed.
In addition, if using mutiprocessing, each shard MUST HAVE THE SAME NUMBER OF ENTRIES after filtering
is applied. We currently do not check for this, but your program may hang if the shards are uneven!
Notice that a few arguments are different from the AudioToBPEDataset; for example, shuffle (bool) has been
replaced by shuffle_n (int).
Additionally, please note that the len() of this DataLayer is assumed to be the length of the manifest
after filtering. An incorrect manifest length may lead to some DataLoader issues down the line.
Args:
audio_tar_filepaths: Either a list of audio tarball filepaths, or a
string (can be brace-expandable).
manifest_filepath (str): Path to the manifest.
tokenizer (TokenizerSpec): Either a Word Piece Encoding tokenizer (BERT),
or a Sentence Piece Encoding tokenizer (BPE). The CTC blank
symbol is automatically added later for models using ctc.
sample_rate (int): Sample rate to resample loaded audio to
int_values (bool): If true, load samples as 32-bit integers. Defauts to False.
augmentor (nemo.collections.asr.parts.perturb.AudioAugmentor): An AudioAugmentor
object used to augment loaded audio
shuffle_n (int): How many samples to look ahead and load to be shuffled.
See WebDataset documentation for more details.
Defaults to 0.
min_duration (float): Dataset parameter.
All training files which have a duration less than min_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to 0.1.
max_duration (float): Dataset parameter.
All training files which have a duration more than max_duration
are dropped. Note: Duration is read from the manifest JSON.
Defaults to None.
max_utts (int): Limit number of utterances. 0 means no maximum.
trim (bool): Whether to use trim silence from beginning and end
of audio signal using librosa.effects.trim().
Defaults to False.
pad_id (id): Token used to pad when collating samples in batches.
If this is None, pads using 0s.
Defaults to None.
shard_strategy (str): Tarred dataset shard distribution strategy chosen as a str value during ddp.
- `scatter`: The default shard strategy applied by WebDataset, where each node gets
a unique set of shards, which are permanently pre-allocated and never changed at runtime.
- `replicate`: Optional shard strategy, where each node gets all of the set of shards
available in the tarred dataset, which are permanently pre-allocated and never changed at runtime.
The benefit of replication is that it allows each node to sample data points from the entire
dataset independently of other nodes, and reduces dependence on value of `shuffle_n`.
Note: Replicated strategy allows every node to sample the entire set of available tarfiles,
and therefore more than one node may sample the same tarfile, and even sample the same
data points! As such, there is no assured guarantee that all samples in the dataset will be
sampled at least once during 1 epoch.
global_rank (int): Worker rank, used for partitioning shards. Defaults to 0.
world_size (int): Total number of processes, used for partitioning shards. Defaults to 0.
"""
def __init__(
self,
audio_tar_filepaths: Union[str, List[str]],
manifest_filepath: str,
tokenizer: 'nemo.collections.common.tokenizers.TokenizerSpec',
sample_rate: int,
int_values: bool = False,
augmentor: Optional['nemo.collections.asr.parts.perturb.AudioAugmentor'] = None,
shuffle_n: int = 0,
min_duration: Optional[float] = None,
max_duration: Optional[float] = None,
max_utts: int = 0,
trim: bool = False,
add_misc: bool = False,
use_start_end_token: bool = True,
shard_strategy: str = "scatter",
global_rank: int = 0,
world_size: int = 0,
):
if use_start_end_token and hasattr(tokenizer, 'bos_token'):
bos_id = tokenizer.bos_id
else:
bos_id = None
if use_start_end_token and hasattr(tokenizer, 'eos_token'):
eos_id = tokenizer.eos_id
else:
eos_id = None
if hasattr(tokenizer, 'pad_token'):
pad_id = tokenizer.pad_id
else:
pad_id = 0
class TokenizerWrapper:
def __init__(self, tokenizer):
self._tokenizer = tokenizer
def __call__(self, text):
t = self._tokenizer.text_to_ids(text)
return t
super().__init__(
audio_tar_filepaths=audio_tar_filepaths,
manifest_filepath=manifest_filepath,
parser=TokenizerWrapper(tokenizer),
sample_rate=sample_rate,
int_values=int_values,
augmentor=augmentor,
shuffle_n=shuffle_n,
min_duration=min_duration,
max_duration=max_duration,
max_utts=max_utts,
trim=trim,
bos_id=bos_id,
eos_id=eos_id,
add_misc=add_misc,
pad_id=pad_id,
shard_strategy=shard_strategy,
global_rank=global_rank,
world_size=world_size,
)