Medix-AI/coqui-tts
Medix-AI
/
coqui-tts
mirror of https://github.com/coqui-ai/TTS.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

data loader fix

This commit is contained in:
Eren Golge 2018-03-22 14:06:33 -07:00
parent 33937f54d0
commit b1beb1f876
1 changed files with 7 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
datasets/LJSpeech.py
View File

@ -97,6 +97,12 @@ class LJSpeechDataset(Dataset):
linear = [self.ap.spectrogram(w).astype('float32') for w in wav]
mel = [self.ap.melspectrogram(w).astype('float32') for w in wav]
mel_lengths = [m.shape[1] for m in mel]
# compute 'stop token' targets
stop_targets = [np.array([0.]*mel_len) for mel_len in mel_lengths]
# PAD stop targets
stop_targets = prepare_stop_target(stop_targets, self.outputs_per_step)
# PAD sequences with largest length of the batch
text = prepare_data(text).astype(np.int32)
@ -106,7 +112,7 @@ class LJSpeechDataset(Dataset):
linear = prepare_tensor(linear)
mel = prepare_tensor(mel)
assert mel.shape[2] == linear.shape[2]
timesteps = mel.shape[2]
timesteps = mel.shape[2]
# PAD with zeros that can be divided by outputs per step
if (timesteps + 1) % self.outputs_per_step != 0:
@ -120,12 +126,6 @@ class LJSpeechDataset(Dataset):
# update mel lengths
mel_lengths = [l+pad_len for l in mel_lengths]
# compute 'stop token' targets
stop_targets = [np.array([0.]*mel_len) for mel_len in mel_lengths]
# PAD stop targets
stop_targets = prepare_stop_target(stop_targets, self.outputs_per_step)
# B x T x D
linear = linear.transpose(0, 2, 1)