|
|
@ -330,11 +330,10 @@ class DataGenerator(object):
|
|
|
|
axis=0)
|
|
|
|
axis=0)
|
|
|
|
masks.append(mask)
|
|
|
|
masks.append(mask)
|
|
|
|
padded_audios = np.array(padded_audios).astype('float32')
|
|
|
|
padded_audios = np.array(padded_audios).astype('float32')
|
|
|
|
|
|
|
|
texts = np.expand_dims(np.array(texts).astype('int32'), axis=-1)
|
|
|
|
if self._is_training:
|
|
|
|
if self._is_training:
|
|
|
|
texts = fluid.create_lod_tensor(
|
|
|
|
texts = fluid.create_lod_tensor(
|
|
|
|
np.array(texts).astype('int32'),
|
|
|
|
texts, recursive_seq_lens=[text_lens], place=self._place)
|
|
|
|
recursive_seq_lens=[text_lens],
|
|
|
|
|
|
|
|
place=self._place)
|
|
|
|
|
|
|
|
audio_lens = np.array(audio_lens).astype('int64').reshape([-1, 1])
|
|
|
|
audio_lens = np.array(audio_lens).astype('int64').reshape([-1, 1])
|
|
|
|
masks = np.array(masks).astype('float32')
|
|
|
|
masks = np.array(masks).astype('float32')
|
|
|
|
return padded_audios, texts, audio_lens, masks
|
|
|
|
return padded_audios, texts, audio_lens, masks
|
|
|
|