def get_datasets(
meta_dir: str,
batch_size: int,
num_workers: int,
fix_len: int = 0,
skip_audio: bool = False) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
# TODO: update this function in general
assert os.path.isdir(meta_dir), '{} is not valid directory path!'
train_file, valid_file = MaestroMeta.frame_file_names[1:]
# load meta file
train_meta = MaestroMeta(os.path.join(meta_dir, train_file))
valid_meta = MaestroMeta(os.path.join(meta_dir, valid_file))
# create dataset
train_dataset = SpeechDataset(train_meta,
fix_len=fix_len,
skip_audio=skip_audio)
valid_dataset = SpeechDataset(valid_meta,
fix_len=fix_len,
skip_audio=skip_audio)
# create data loader
train_loader = SpeechDataLoader(train_dataset,
batch_size=batch_size,
num_workers=num_workers)
valid_loader = SpeechDataLoader(valid_dataset,
batch_size=batch_size,
num_workers=num_workers)
return train_loader, valid_loader
def get_datasets(
meta_dir: str, batch_size: int,
num_workers: int) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
assert os.path.isdir(meta_dir), '{} is not valid directory path!'
train_file, valid_file = LJSpeechMeta.frame_file_names[1:]
# load meta file
train_meta = LJSpeechMeta(os.path.join(meta_dir, train_file))
valid_meta = LJSpeechMeta(os.path.join(meta_dir, valid_file))
# create dataset
train_dataset = SpeechTTSDataset(train_meta)
valid_dataset = SpeechTTSDataset(valid_meta)
# create data loader
train_loader = SpeechDataLoader(train_dataset,
batch_size=batch_size,
is_bucket=True,
num_workers=num_workers,
n_buckets=5)
valid_loader = SpeechDataLoader(valid_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers)
return train_loader, valid_loader
def get_datasets(
meta_dir: str,
batch_size: int,
num_workers: int,
fix_len: int = 0,
audio_mask: bool = False) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
assert os.path.isdir(meta_dir), '{} is not valid directory path!'
train_file, valid_file = MedleyDBMeta.frame_file_names[1:]
# load meta file
train_meta = MedleyDBMeta(os.path.join(meta_dir, train_file))
valid_meta = MedleyDBMeta(os.path.join(meta_dir, valid_file))
# create dataset
train_dataset = SpeechDataset(train_meta,
fix_len=fix_len,
audio_mask=audio_mask)
valid_dataset = SpeechDataset(valid_meta,
fix_len=fix_len,
audio_mask=audio_mask)
# create data loader
train_loader = SpeechDataLoader(train_dataset,
batch_size=batch_size,
num_workers=num_workers,
is_bucket=False)
valid_loader = SpeechDataLoader(valid_dataset,
batch_size=batch_size,
num_workers=num_workers,
is_bucket=False)
return train_loader, valid_loader
def get_concated_datasets(
meta_dir_list: List[str],
batch_size: int,
num_workers: int,
meta_cls_list: List[MetaFrame],
fix_len: int = 0,
skip_audio: bool = False,
sample_rate: int = 44100,
audio_mask: bool = False) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
assert all([os.path.isdir(x) for x in meta_dir_list
]), 'There are not valid directory paths!'.format()
assert len(meta_dir_list) == len(
meta_cls_list
), 'meta_dir_list, meta_cls_list are must have same length!'
# datasets
train_datasets = []
valid_datasets = []
for meta_cls, meta_dir in zip(meta_cls_list, meta_dir_list):
train_file, valid_file = meta_cls.frame_file_names[1:]
# load meta file
train_meta = meta_cls(os.path.join(meta_dir, train_file),
sr=sample_rate)
valid_meta = meta_cls(os.path.join(meta_dir, valid_file),
sr=sample_rate)
# create dataset
train_dataset = AugmentSpeechDataset(train_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask)
valid_dataset = AugmentSpeechDataset(valid_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask)
train_datasets.append(train_dataset)
valid_datasets.append(valid_dataset)
# make concat dataset
train_conc_dataset = ConcatDataset(train_datasets)
valid_conc_dataset = ConcatDataset(valid_datasets)
# create data loader
train_loader = SpeechDataLoader(train_conc_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers,
skip_last_bucket=False)
valid_loader = SpeechDataLoader(valid_conc_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers,
skip_last_bucket=False)
return train_loader, valid_loader
def get_datasets(
meta_dir: str,
batch_size: int,
num_workers: int,
meta_cls: MetaFrame = VoiceBankMeta,
fix_len: int = 0,
skip_audio: bool = False,
audio_mask: bool = False,
is_audioset: bool = False
) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
assert os.path.isdir(meta_dir), '{} is not valid directory path!'
train_file, valid_file = meta_cls.frame_file_names[1:]
# load meta file
train_meta = meta_cls(os.path.join(meta_dir, train_file))
valid_meta = meta_cls(os.path.join(meta_dir, valid_file))
# create dataset
train_dataset = AugmentSpeechDataset(train_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask,
is_audioset=is_audioset)
valid_dataset = AugmentSpeechDataset(valid_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask,
is_audioset=is_audioset)
# create data loader
train_loader = SpeechDataLoader(train_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers,
skip_last_bucket=False)
valid_loader = SpeechDataLoader(valid_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers,
skip_last_bucket=False)
return train_loader, valid_loader
def get_datasets(
meta_dir: str,
batch_size: int,
num_workers: int,
fix_len: int = 0,
skip_audio: bool = False,
audio_mask: bool = False,
skip_last_bucket: bool = True,
n_buckets: int = 10,
extra_features: List[Tuple[str, Callable]] = None
) -> Tuple[SpeechDataLoader, SpeechDataLoader]:
assert os.path.isdir(meta_dir), '{} is not valid directory path!'
train_file, valid_file = LibriTTSMeta.frame_file_names[1:]
# load meta file
train_meta = LibriTTSMeta(os.path.join(meta_dir, train_file))
valid_meta = LibriTTSMeta(os.path.join(meta_dir, valid_file))
# create dataset
train_dataset = SpeechDataset(train_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask,
extra_features=extra_features)
valid_dataset = SpeechDataset(valid_meta,
fix_len=fix_len,
skip_audio=skip_audio,
audio_mask=audio_mask,
extra_features=extra_features)
# create data loader
train_loader = SpeechDataLoader(train_dataset,
batch_size=batch_size,
n_buckets=n_buckets,
num_workers=num_workers,
skip_last_bucket=skip_last_bucket)
valid_loader = SpeechDataLoader(valid_dataset,
batch_size=batch_size,
is_bucket=False,
num_workers=num_workers)
return train_loader, valid_loader