wassname/TTS
1
0
Fork 0
mirror of https://github.com/wassname/TTS.git synced 2026-09-09 11:16:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

style update #3

Browse Source
This commit is contained in:
Eren Gölge
2021-04-09 01:17:15 +02:00
parent 18d9ec8036
commit 87ee6ceb57
30 changed files with 717 additions and 682 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test_loader.py
+18 -18
View File
@@ -12,11 +12,11 @@ from TTS.tts.datasets.preprocess import ljspeech
from TTS.utils.audio import AudioProcessor
from TTS.utils.io import load_config
#pylint: disable=unused-variable
# pylint: disable=unused-variable
OUTPATH = os.path.join(get_tests_output_path(), "loader_tests/")
os.makedirs(OUTPATH, exist_ok=True)
c = load_config(os.path.join(get_tests_input_path(), 'test_config.json'))
c = load_config(os.path.join(get_tests_input_path(), "test_config.json"))
ok_ljspeech = os.path.exists(c.data_path)
DATA_EXIST = True
@@ -33,25 +33,27 @@ class TestTTSDataset(unittest.TestCase):
self.ap = AudioProcessor(**c.audio)
def _create_dataloader(self, batch_size, r, bgs):
items = ljspeech(c.data_path, 'metadata.csv')
items = ljspeech(c.data_path, "metadata.csv")
dataset = TTSDataset.MyDataset(
r,
c.text_cleaner,
compute_linear_spec=True,
ap=self.ap,
meta_data=items,
tp=c.characters if 'characters' in c.keys() else None,
tp=c.characters if "characters" in c.keys() else None,
batch_group_size=bgs,
min_seq_len=c.min_seq_len,
max_seq_len=float("inf"),
use_phonemes=False)
use_phonemes=False,
)
dataloader = DataLoader(
dataset,
batch_size=batch_size,
shuffle=False,
collate_fn=dataset.collate_fn,
drop_last=True,
num_workers=c.num_loader_workers)
num_workers=c.num_loader_workers,
)
return dataloader, dataset
def test_loader(self):
@@ -72,18 +74,17 @@ class TestTTSDataset(unittest.TestCase):
neg_values = text_input[text_input < 0]
check_count = len(neg_values)
assert check_count == 0, \
" !! Negative values in text_input: {}".format(check_count)
assert check_count == 0, " !! Negative values in text_input: {}".format(check_count)
# TODO: more assertion here
assert isinstance(speaker_name[0], str)
assert linear_input.shape[0] == c.batch_size
assert linear_input.shape[2] == self.ap.fft_size // 2 + 1
assert mel_input.shape[0] == c.batch_size
assert mel_input.shape[2] == c.audio['num_mels']
assert mel_input.shape[2] == c.audio["num_mels"]
# check normalization ranges
if self.ap.symmetric_norm:
assert mel_input.max() <= self.ap.max_norm
assert mel_input.min() >= -self.ap.max_norm #pylint: disable=invalid-unary-operand-type
assert mel_input.min() >= -self.ap.max_norm # pylint: disable=invalid-unary-operand-type
assert mel_input.min() < 0
else:
assert mel_input.max() <= self.ap.max_norm
@@ -134,7 +135,7 @@ class TestTTSDataset(unittest.TestCase):
# check mel_spec consistency
wav = np.asarray(self.ap.load_wav(item_idx[0]), dtype=np.float32)
mel = self.ap.melspectrogram(wav).astype('float32')
mel = self.ap.melspectrogram(wav).astype("float32")
mel = torch.FloatTensor(mel).contiguous()
mel_dl = mel_input[0]
# NOTE: Below needs to check == 0 but due to an unknown reason
@@ -145,15 +146,14 @@ class TestTTSDataset(unittest.TestCase):
# check mel-spec correctness
mel_spec = mel_input[0].cpu().numpy()
wav = self.ap.inv_melspectrogram(mel_spec.T)
self.ap.save_wav(wav, OUTPATH + '/mel_inv_dataloader.wav')
shutil.copy(item_idx[0], OUTPATH + '/mel_target_dataloader.wav')
self.ap.save_wav(wav, OUTPATH + "/mel_inv_dataloader.wav")
shutil.copy(item_idx[0], OUTPATH + "/mel_target_dataloader.wav")
# check linear-spec
linear_spec = linear_input[0].cpu().numpy()
wav = self.ap.inv_spectrogram(linear_spec.T)
self.ap.save_wav(wav, OUTPATH + '/linear_inv_dataloader.wav')
shutil.copy(item_idx[0],
OUTPATH + '/linear_target_dataloader.wav')
self.ap.save_wav(wav, OUTPATH + "/linear_inv_dataloader.wav")
shutil.copy(item_idx[0], OUTPATH + "/linear_target_dataloader.wav")
# check the last time step to be zero padded
assert linear_input[0, -1].sum() != 0
@@ -202,8 +202,8 @@ class TestTTSDataset(unittest.TestCase):
# check the second itme in the batch
assert linear_input[1 - idx, -1].sum() == 0
assert mel_input[1 - idx, -1].sum() == 0
assert stop_target[1, mel_lengths[1]-1] == 1
assert stop_target[1, mel_lengths[1]:].sum() == 0
assert stop_target[1, mel_lengths[1] - 1] == 1
assert stop_target[1, mel_lengths[1] :].sum() == 0
assert len(mel_lengths.shape) == 1
# check batch zero-frame conditions (zero-frame disabled)