Fix tokenizer for punc only (#1717)

coqui-ai · Jul 6, 2022 · 5cef6fa · 5cef6fa
1 parent 9e00e31
commit 5cef6fa
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 1 deletion.
diff --git a/TTS/tts/utils/text/punctuation.py b/TTS/tts/utils/text/punctuation.py
@@ -137,7 +137,7 @@ def _restore(cls, text, puncs, num): # pylint: disable=too-many-return-statemen
 
  # nothing have been phonemized, returns the puncs alone
  if not text:
- return ["".join(m.mark for m in puncs)]
+ return ["".join(m.punc for m in puncs)]
 
  current = puncs[0]
 

diff --git a/tests/text_tests/test_tokenizer.py b/tests/text_tests/test_tokenizer.py
@@ -30,6 +30,13 @@ def test_text_to_ids_phonemes(self):
  test_hat = self.tokenizer_ph.ids_to_text(ids)
  self.assertEqual(text_ph, test_hat)
 
+ def test_text_to_ids_phonemes_punctuation(self):
+ text = "..."
+ text_ph = self.ph.phonemize(text, separator="")
+ ids = self.tokenizer_ph.text_to_ids(text)
+ test_hat = self.tokenizer_ph.ids_to_text(ids)
+ self.assertEqual(text_ph, test_hat)
+
  def test_text_to_ids_phonemes_with_eos_bos(self):
  text = "Bu bir Örnek."
  self.tokenizer_ph.use_eos_bos = True