|
|
|
@ -194,7 +194,7 @@ class SpeechCollatorBase():
|
|
|
|
|
texts = []
|
|
|
|
|
text_lens = []
|
|
|
|
|
utts = []
|
|
|
|
|
tids = [] # tokenids
|
|
|
|
|
tids = [] # tokenids
|
|
|
|
|
|
|
|
|
|
for idx, item in enumerate(batch):
|
|
|
|
|
utts.append(item['utt'])
|
|
|
|
@ -354,7 +354,7 @@ class TripletSpeechCollator(SpeechCollator):
|
|
|
|
|
|
|
|
|
|
tokens = [[], []]
|
|
|
|
|
for idx, text in enumerate([translation, transcription]):
|
|
|
|
|
tokens[idx] = tokenids(text, self.keep_transcription_text)
|
|
|
|
|
tokens[idx] = _tokenids(text, self.keep_transcription_text)
|
|
|
|
|
|
|
|
|
|
translation_text.append(tokens[0])
|
|
|
|
|
translation_text_lens.append(tokens[0].shape[0])
|
|
|
|
|