def test_SpecAuc(apply_time_warp, apply_freq_mask, apply_time_mask):
if not apply_time_warp and not apply_time_mask and not apply_freq_mask:
with pytest.raises(ValueError):
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
)
else:
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
)
x = torch.randn(2, 1000, 80)
specaug(x)
def test_SpecAuc_repr(
apply_time_warp,
apply_freq_mask,
apply_time_mask,
time_mask_width_range,
time_mask_width_ratio_range,
):
if (
(not apply_time_warp and not apply_time_mask and not apply_freq_mask)
or (
apply_time_mask
and time_mask_width_range is None
and time_mask_width_ratio_range is None
)
or (
apply_time_mask
and time_mask_width_range is not None
and time_mask_width_ratio_range is not None
)
):
return
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
time_mask_width_range=time_mask_width_range,
time_mask_width_ratio_range=time_mask_width_ratio_range,
)
print(specaug)
def __init__(self, input_size: int, vocab_size: int,
token_list: Union[Tuple[str, ...],
List[str]], device, config):
super().__init__()
self.sos = vocab_size - 1
self.eos = vocab_size - 1
self.vocab_size = vocab_size
self.ignore_id = config.ignore_id
self.ctc_weight = config.mtlalpha
self.token_list = config.char_list.copy()
self.specaug = SpecAug() if config.specaug else None
self.normalize = UtteranceMVN() if config.normalize else None
print(self.specaug, self.normalize)
self.frontend = CustomFrontend(fs=SAMPLE_RATE,
n_fft=512,
normalized=True,
hop_length=int(0.01 * SAMPLE_RATE),
win_length=int(0.03 * SAMPLE_RATE),
n_mels=80)
self.model = E2E(input_size, vocab_size, config, device)
def test_SpecAuc_repr(apply_time_warp, apply_freq_mask, apply_time_mask):
if not apply_time_warp and not apply_time_mask and not apply_freq_mask:
return
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
)
print(specaug)
def __init__(self, sos, eos, is_spec_aug=False, dtype='FLOAT16'):
'''
Args:
sos: sos token id
eos: eos token id
is_spec_aug(bool): Whether to use feature spec augment
'''
self.sos = sos
self.eos = eos
self.is_spec_aug = is_spec_aug
self.spec_aug = SpecAug()
self.dtype = dtype
def test_SpecAuc(
apply_time_warp,
apply_freq_mask,
apply_time_mask,
time_mask_width_range,
time_mask_width_ratio_range,
):
if (
(not apply_time_warp and not apply_time_mask and not apply_freq_mask)
or (
apply_time_mask
and time_mask_width_range is None
and time_mask_width_ratio_range is None
)
or (
apply_time_mask
and time_mask_width_range is not None
and time_mask_width_ratio_range is not None
)
):
with pytest.raises(ValueError):
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
time_mask_width_range=time_mask_width_range,
time_mask_width_ratio_range=time_mask_width_ratio_range,
)
else:
specaug = SpecAug(
apply_time_warp=apply_time_warp,
apply_freq_mask=apply_freq_mask,
apply_time_mask=apply_time_mask,
time_mask_width_range=time_mask_width_range,
time_mask_width_ratio_range=time_mask_width_ratio_range,
)
x = torch.randn(2, 1000, 80)
specaug(x)
def get_specaug():
return SpecAug(
apply_time_warp=True,
apply_freq_mask=True,
apply_time_mask=False,
)