set use_phonemes false

TTS/tts/datasets/TTSDataset.py

@@ -25,7 +25,7 @@ class MyDataset(Dataset):
         batch_group_size=0,
         min_seq_len=0,
         max_seq_len=float("inf"),
-        use_phonemes=True,
+        use_phonemes=False,
         phoneme_cache_path=None,
         phoneme_language="en-us",
         enable_eos_bos=False,

TTS/tts/utils/text/__init__.py

@@ -4,8 +4,8 @@ import re
 
 from packaging import version
 
-from TTS.tts.utils.text.chinese_mandarin.phonemizer import chinese_text_to_phonemes
 from TTS.tts.utils.text import cleaners
+from TTS.tts.utils.text.chinese_mandarin.phonemizer import chinese_text_to_phonemes
 from TTS.tts.utils.text.symbols import _bos, _eos, _punctuations, make_symbols, phonemes, symbols
 
 # pylint: disable=unnecessary-comprehension
@@ -27,12 +27,12 @@ PHONEME_PUNCTUATION_PATTERN = r"[" + _punctuations.replace(" ", "") + "]+"
 
 def text2phone(text, language):
     """Convert graphemes to phonemes.
-        Parameters:
-                text (str): text to phonemize
-                language (str): language of the text
-        Returns:
-                ph (str): phonemes as a string seperated by "|"
-                        ph = "ɪ|g|ˈ|z|æ|m|p|ə|l"
+    Parameters:
+            text (str): text to phonemize
+            language (str): language of the text
+    Returns:
+            ph (str): phonemes as a string seperated by "|"
+                    ph = "ɪ|g|ˈ|z|æ|m|p|ə|l"
     """
 
     # TO REVIEW : How to have a good implementation for this?

TTS/utils/manage.py

@@ -102,8 +102,10 @@ class ModelManager(object):
         output_model_path = os.path.join(output_path, "model_file.pth.tar")
         output_config_path = os.path.join(output_path, "config.json")
         # NOTE : band-aid for removing phoneme support
-        if 'needs_phonemizer' in model_item and model_item['needs_phonemizer']:
-            raise RuntimeError(' [!] Use 🐸TTS <= v0.0.13 for this model. Current version does not support phoneme based models.')
+        if "needs_phonemizer" in model_item and model_item["needs_phonemizer"]:
+            raise RuntimeError(
+                " [!] Use 🐸TTS <= v0.0.13 for this model. Current version does not support phoneme based models."
+            )
         if os.path.exists(output_path):
             print(f" > {model_name} is already downloaded.")
         else:

tests/tts_tests/test_align_tts_train.py

@@ -15,7 +15,7 @@ config = AlignTTSConfig(
     num_loader_workers=0,
     num_val_loader_workers=0,
     text_cleaner="english_cleaners",
-    use_phonemes=True,
+    use_phonemes=False,
     phoneme_language="en-us",
     phoneme_cache_path=os.path.join(get_tests_output_path(), "train_outputs/phoneme_cache/"),
     run_eval=True,

tests/tts_tests/test_glow_tts_train.py

@@ -15,7 +15,7 @@ config = GlowTTSConfig(
     num_loader_workers=0,
     num_val_loader_workers=0,
     text_cleaner="english_cleaners",
-    use_phonemes=True,
+    use_phonemes=False,
     phoneme_language="en-us",
     phoneme_cache_path=os.path.join(get_tests_output_path(), "train_outputs/phoneme_cache/"),
     run_eval=True,

tests/tts_tests/test_speedy_speech_train.py

@@ -15,7 +15,7 @@ config = SpeedySpeechConfig(
     num_loader_workers=0,
     num_val_loader_workers=0,
     text_cleaner="english_cleaners",
-    use_phonemes=True,
+    use_phonemes=False,
     phoneme_language="en-us",
     phoneme_cache_path=os.path.join(get_tests_output_path(), "train_outputs/phoneme_cache/"),
     run_eval=True,

tests/tts_tests/test_tacotron2_train.py

@@ -16,7 +16,7 @@ config = Tacotron2Config(
     num_loader_workers=0,
     num_val_loader_workers=0,
     text_cleaner="english_cleaners",
-    use_phonemes=True,
+    use_phonemes=False,
     phoneme_language="en-us",
     phoneme_cache_path=os.path.join(get_tests_output_path(), "train_outputs/phoneme_cache/"),
     run_eval=True,

tests/tts_tests/test_tacotron_train.py

@@ -15,7 +15,7 @@ config = TacotronConfig(
     num_loader_workers=0,
     num_val_loader_workers=0,
     text_cleaner="english_cleaners",
-    use_phonemes=True,
+    use_phonemes=False,
     phoneme_language="en-us",
     phoneme_cache_path=os.path.join(get_tests_output_path(), "train_outputs/phoneme_cache/"),
     run_eval=True,