phoonnx 0.0.1a1__py3-none-any.whl → 0.0.2a2__py3-none-any.whl

phoonnx/config.py

@@ -6,7 +6,6 @@ from phoonnx.phoneme_ids import (load_phoneme_ids, BlankBetween,
                                  DEFAULT_BLANK_WORD_TOKEN, DEFAULT_BLANK_TOKEN,
                                  DEFAULT_PAD_TOKEN, DEFAULT_BOS_TOKEN, DEFAULT_EOS_TOKEN)
 
-
 DEFAULT_NOISE_SCALE = 0.667
 DEFAULT_LENGTH_SCALE = 1.0
 DEFAULT_NOISE_W_SCALE = 0.8
@@ -22,6 +21,8 @@ class Alphabet(str, Enum):
     UNICODE = "unicode"
     IPA = "ipa"
     ARPA = "arpa" # en
+    SAMPA = "sampa"
+    XSAMPA = "x-sampa"
     HANGUL = "hangul" # ko
     KANA = "kana" # ja
     HIRA = "hira" # ja
@@ -32,6 +33,7 @@ class Alphabet(str, Enum):
     ERAAB = "eraab" # fa
     COTOVIA = "cotovia" # gl
     HANZI = "hanzi" # zh
+    MANTOQ = "mantoq" # ar
 
 
 

phoonnx/phonemizers/ar.py

@@ -5,7 +5,7 @@ from phoonnx.config import Alphabet
 class MantoqPhonemizer(BasePhonemizer):
 
     def __init__(self):
-        super().__init__(Alphabet.IPA)
+        super().__init__(Alphabet.MANTOQ)
 
     @classmethod
     def get_lang(cls, target_lang: str) -> str:

phoonnx/phonemizers/gl.py

@@ -12,6 +12,57 @@ class CotoviaError(Exception):
     pass
 
 
+COTOVIA2IPA = {
+    "pau": " ",
+    "a": "a",
+    "E": "ɛ",
+    "e": "e",
+    "i": "i",
+    "j": "j",
+    "O": "ɔ",
+    "o": "o",
+    "u": "u",
+    "w": "w",
+    "p": "p",
+    "b": "b",
+    "B": "β",
+    "t": "t",
+    "d": "d",
+    "D": "ð",
+    "k": "k",
+    "g": "g",
+    "G": "ɣ",
+    "f": "f",
+    "T": "θ",
+    "s": "s",
+    "S": "ʃ",
+    "tS": "tʃ",
+    "m": "m",
+    "n": "n",
+    "J": "ɲ",
+    "N": "ŋ",
+    "l": "l",
+    "Z": "ʎ",
+    "jj": "ʎ",
+    "L": "ʎ",
+    "r": "ɾ",
+    "rr": "r",
+    "X": "x"
+}
+
+
+def cotovia2ipa(text: str) -> str:
+    """
+    Converts a string of Cotovía phonemes to IPA.
+    """
+    # Sort the dictionary keys by length in descending order to handle multi-character phonemes first
+    sorted_cotovia_keys = sorted(COTOVIA2IPA.keys(), key=len, reverse=True)
+    ipa_str = text
+    for cotovia_char in sorted_cotovia_keys:
+        ipa_str = ipa_str.replace(cotovia_char, COTOVIA2IPA[cotovia_char])
+    return ipa_str
+
+
 class CotoviaPhonemizer(BasePhonemizer):
     """
     A phonemizer class that uses the Cotovia TTS binary to convert text into phonemes.
@@ -19,7 +70,7 @@ class CotoviaPhonemizer(BasePhonemizer):
     regular expression transformations to clean and normalize the phonetic representation.
     """
 
-    def __init__(self, cotovia_bin_path: Optional[str] = None):
+    def __init__(self, cotovia_bin_path: Optional[str] = None, alphabet: Alphabet = Alphabet.IPA):
         """
         Initializes the CotoviaPhonemizer.
 
@@ -31,7 +82,7 @@ class CotoviaPhonemizer(BasePhonemizer):
         if not os.path.exists(self.cotovia_bin):
             raise FileNotFoundError(f"Cotovia binary not found at {self.cotovia_bin}. "
                                     "Please ensure it's installed or provide the correct path.")
-        super().__init__(Alphabet.COTOVIA)
+        super().__init__(alphabet)
 
     @classmethod
     def get_lang(cls, target_lang: str) -> str:
@@ -127,6 +178,8 @@ class CotoviaPhonemizer(BasePhonemizer):
         # substitute ' ( text )' to ', text,'
         str_ext = re.sub(r"(\w+)\s*\(\s*([^\(\)]*?)\s*\)", r"\1, \\2,", str_ext)
 
+        if self.alphabet == Alphabet.IPA:
+            return cotovia2ipa(str_ext)
         return str_ext
 
 
@@ -138,5 +191,5 @@ if __name__ == "__main__":
     lang = "gl"
     text_gl = "Este é un sistema de conversión de texto a voz en lingua galega baseado en redes neuronais artificiais. Ten en conta que as funcionalidades incluídas nesta páxina ofrécense unicamente con fins de demostración. Se tes algún comentario, suxestión ou detectas algún problema durante a demostración, ponte en contacto connosco."
     print(f"\n--- Getting phonemes for '{text_gl}' (Cotovia) ---")
-    phonemes_cotovia = cotovia.phonemize(text_gl, lang)
+    phonemes_cotovia = cotovia.phonemize_string(text_gl, lang)
     print(f"  Cotovia Phonemes: {phonemes_cotovia}")

phoonnx/util.py

@@ -567,15 +567,18 @@ def _normalize_units(text: str, full_lang: str) -> str:
             symbolic_pattern = re.compile(number_pattern_str + r"\s*(" + symbolic_pattern_str + r")", re.IGNORECASE)
 
             def replace_symbolic(match):
-                number_str = match.group(1)
+                number = match.group(1)
                 # Remove thousands separator and replace decimal separator for parsing
-                number = number_str.replace(thousands_separator, "").replace(decimal_separator, ".")
+                if thousands_separator in number and decimal_separator in number:
+                    number = number.replace(thousands_separator, "").replace(decimal_separator, ".")
+                elif decimal_separator != "." and decimal_separator in number:
+                    number = number.replace(decimal_separator, ".")
                 unit_symbol = match.group(2)
                 unit_word = symbolic_units[unit_symbol]
                 try:
-                    return f"{pronounce_number(float(number), full_lang)} {unit_word}"
+                    return f"{pronounce_number(float(number) if '.' in number else int(number), full_lang)} {unit_word}"
                 except Exception as e:
-                    LOG.error(f"Failed to pronounce number with unit: {number_str}{unit_symbol} - ({e})")
+                    LOG.error(f"Failed to pronounce number with unit: {number}{unit_symbol} - ({e})")
                     return match.group(0)
             text = symbolic_pattern.sub(replace_symbolic, text)
 
@@ -588,12 +591,15 @@ def _normalize_units(text: str, full_lang: str) -> str:
                                               re.IGNORECASE)
 
             def replace_alphanumeric(match):
-                number_str = match.group(1)
+                number = match.group(1)
                 # Remove thousands separator and replace decimal separator for parsing
-                number = number_str.replace(thousands_separator, "").replace(decimal_separator, ".")
+                if thousands_separator in number and decimal_separator in number:
+                    number = number.replace(thousands_separator, "").replace(decimal_separator, ".")
+                elif decimal_separator != "." and decimal_separator in number:
+                    number = number.replace(decimal_separator, ".")
                 unit_symbol = match.group(2)
                 unit_word = alphanumeric_units[unit_symbol]
-                return f"{pronounce_number(float(number), full_lang)} {unit_word}"
+                return f"{pronounce_number(float(number) if '.' in number else int(number), full_lang)} {unit_word}"
 
             text = alphanumeric_pattern.sub(replace_alphanumeric, text)
     return text
@@ -667,7 +673,8 @@ if __name__ == "__main__":
 
     # General normalization examples
     print("General English example: " + normalize('I\'m Dr. Prof. 3/3 0.5% of 12345€, 5ft, and 10kg', 'en'))
-    print(f"General Portuguese example: {normalize('Dr. Prof. 3/3 0.5% de 12345€, 5m, e 10kg', 'pt')}")
+    print(f"Word Salad Portuguese (Dr. Prof. 3/3 0,5% de 12345€, 5m, e 10kg): {normalize('Dr. Prof. 3/3 0,5% de 12345€, 5m, e 10kg', 'pt')}")
+    print(f"Word Salad Portuguese (Dr. Prof. 3/3 0.5% de 12345€, 5m, e 10kg): {normalize('Dr. Prof. 3/3 0.5% de 12345€, 5m, e 10kg', 'pt')}")
 
     # Portuguese examples with comma decimal separator
     print("\n--- Portuguese Decimal Separator Examples ---")
@@ -691,10 +698,10 @@ if __name__ == "__main__":
 
     # Portuguese dates and times
     print("\n--- Portuguese Date & Time Examples ---")
-    print(f"Portuguese date (DMY format): {normalize('A data é 03/08/2025', 'pt')}")
-    print(f"Portuguese ambiguous date (DMY assumed): {normalize('O relatório é para 15/05/2025', 'pt')}")
-    print(f"Portuguese date with dashes: {normalize('O evento é no dia 25-10-2024', 'pt')}")
-    print(f"Portuguese military time: {normalize('O encontro é às 14h30', 'pt')}")
+    print(f"Portuguese date (A data é 03/08/2025): {normalize('A data é 03/08/2025', 'pt')}")
+    print(f"Portuguese ambiguous date (O relatório é para 15/05/2025): {normalize('O relatório é para 15/05/2025', 'pt')}")
+    print(f"Portuguese date with dashes (O evento é no dia 25-10-2024): {normalize('O evento é no dia 25-10-2024', 'pt')}")
+    print(f"Portuguese military time (O encontro é às 14h30): {normalize('O encontro é às 14h30', 'pt')}")
 
     # Other examples
     print(f"\n--- Other Examples ---")

phoonnx/version.py

@@ -1,6 +1,6 @@
 # START_VERSION_BLOCK
 VERSION_MAJOR = 0
 VERSION_MINOR = 0
-VERSION_BUILD = 1
-VERSION_ALPHA = 1
+VERSION_BUILD = 2
+VERSION_ALPHA = 2
 # END_VERSION_BLOCK

{phoonnx-0.0.1a1.dist-info → phoonnx-0.0.2a2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.0.1a1
+Version: 0.0.2a2
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com
@@ -8,7 +8,7 @@ Requires-Dist: numpy
 Requires-Dist: onnxruntime
 Requires-Dist: quebra-frases
 Requires-Dist: langcodes
-Requires-Dist: ovos-number-parser>=0.3.3a1
+Requires-Dist: ovos-number-parser>=0.4.0
 Requires-Dist: ovos-date-parser>=0.6.4a1
 Provides-Extra: aa
 Requires-Dist: epitran; extra == "aa"
@@ -214,6 +214,12 @@ Provides-Extra: tpi
 Requires-Dist: epitran; extra == "tpi"
 Provides-Extra: tr
 Requires-Dist: epitran; extra == "tr"
+Provides-Extra: train
+Requires-Dist: cython<1,>=0.29.0; extra == "train"
+Requires-Dist: librosa<1,>=0.9.2; extra == "train"
+Requires-Dist: numpy<2,>=1.19.0; extra == "train"
+Requires-Dist: pytorch-lightning<2.0; extra == "train"
+Requires-Dist: torch<2,>=1.11.0; extra == "train"
 Provides-Extra: uew
 Requires-Dist: epitran; extra == "uew"
 Provides-Extra: ug

{phoonnx-0.0.1a1.dist-info → phoonnx-0.0.2a2.dist-info}/RECORD

@@ -1,19 +1,19 @@
 phoonnx/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-phoonnx/config.py,sha256=bO7dx2tfLotkohict3UKlCEVm-BRFB1feYYR1HarUkk,19382
+phoonnx/config.py,sha256=81H34oPG2BaiOA6UM1KapoT341n068LqRprKb5ER6mY,19451
 phoonnx/phoneme_ids.py,sha256=FiNgZwV6naEsBh6XwFLh3_FyOgPiCsK9qo7S0v-CmI4,13667
-phoonnx/util.py,sha256=B2PmkxQ14BJvcM7SmyZtRu04oxuA8pfVTqPlP2PpuTM,25121
-phoonnx/version.py,sha256=bNAgeuunrAuqqTHQ6aL8PuYgpU_pmqDADvHLxvFez88,114
+phoonnx/util.py,sha256=XSjFEoqSFcujFTHxednacgC9GrSYyF-Il5L6Utmxmu4,25909
+phoonnx/version.py,sha256=E5lPoALEg4bqb4bmnVYnseCgozkzjZGnhpUom_OlcQI,114
 phoonnx/voice.py,sha256=FR_LafK1vSi_anPERJjZBuH3Bb9vUIof0MAW6TnALlA,20024
 phoonnx/locale/ca/phonetic_spellings.txt,sha256=igv3t7jxLSRE5GHsdn57HOpxiWNcEmECPql6m02wbO0,47
 phoonnx/locale/en/phonetic_spellings.txt,sha256=xGQlWOABLzbttpQvopl9CU-NnwEJRqKx8iuylsdUoQA,27
 phoonnx/locale/gl/phonetic_spellings.txt,sha256=igv3t7jxLSRE5GHsdn57HOpxiWNcEmECPql6m02wbO0,47
 phoonnx/locale/pt/phonetic_spellings.txt,sha256=KntS8QMynEJ5A3Clvcjq4qlmL-ThSbhfD6v0nKSrlqs,49
 phoonnx/phonemizers/__init__.py,sha256=QGBZk0QUgJdg2MwUWY9Kpk6ucwrEJYtHb07YcNvXCV4,1647
-phoonnx/phonemizers/ar.py,sha256=W3wJ_GI8gV3qb9fng5xeazX8ZlY-aNSywATTdxryiZg,1257
+phoonnx/phonemizers/ar.py,sha256=rPAMGPlyXOlKY99IoHe_vWcLllepiG0YFUbOBia075E,1260
 phoonnx/phonemizers/base.py,sha256=yPg6-dvscYpl3rR3JEULG1PRF-i8DWC_C3HAZGLbxOo,7648
 phoonnx/phonemizers/en.py,sha256=N2SVoVhplQao7Ej5TXbxJU-YkAgkY0Fr9iYBFnsjFSE,9271
 phoonnx/phonemizers/fa.py,sha256=d_DZM2wqomf4gcRH_rFcNA3VkQWKHru8vwBwaNG8Ll8,1452
-phoonnx/phonemizers/gl.py,sha256=05DX_9lQa3Uws3RPcWPtw0nXX6Vo3bybrZkpMErguTo,5529
+phoonnx/phonemizers/gl.py,sha256=jEFKJJViHufZtB7lGNwWQCdWGiNKDCVZ_GRYXTaw_2c,6614
 phoonnx/phonemizers/he.py,sha256=KbRI3XRZa8UtJdNWmn_fd-t5lmFSIp4Mw8UgcO5l-Po,2211
 phoonnx/phonemizers/ja.py,sha256=Xojsrt715ihnIiEk9K6giYqDo9Iykw-SHfIidrHtHSU,3834
 phoonnx/phonemizers/ko.py,sha256=kwWoOFqanCB8kv2JRx17A0hP78P1wbXlX6e8VBn1ezQ,2989
@@ -70,7 +70,7 @@ phoonnx_train/vits/attentions.py,sha256=yc_ViF8zR8z68DzphmVVVn27f9xK_5wi8S4ITLXV
 phoonnx_train/vits/commons.py,sha256=JsD8CdZ3ZcYYubYhw8So5hICBziFlCrKLrv1lMDRCDM,4645
 phoonnx_train/vits/config.py,sha256=oSuUIhw9Am7BQ5JwDgtCO-P1zRyN7nPgR-U1XuncJls,10789
 phoonnx_train/vits/dataset.py,sha256=DLLGSCkn3GF9uktoTprH1ERblZ18GO6-QsClQKWa98o,6804
-phoonnx_train/vits/lightning.py,sha256=iJJNAT1-jm0wYvTfBy9T_eZapDN62Pny9YdHEpnApso,12298
+phoonnx_train/vits/lightning.py,sha256=ZBuSIiJ7EUU1Za2V8Uh6-_HGGRW_qwpXLLs1cEDirHA,12301
 phoonnx_train/vits/losses.py,sha256=j-uINhBcYxVXFvFutiewQpTuw-qF-J6M6hdJVeOKqNE,1401
 phoonnx_train/vits/mel_processing.py,sha256=huIjbQgewSmM39hdzRZvZUCI7fTNSMmLcAv3f8zYb8k,3956
 phoonnx_train/vits/models.py,sha256=9PziprRtkdBQ6AowFe1vG4QTCk02By-LDS9W8EtZGvE,24303
@@ -80,7 +80,7 @@ phoonnx_train/vits/utils.py,sha256=exiyrtPHbnnGvcHWSbaH9-gR6srH5ZPHlKiqV2IHUrQ,4
 phoonnx_train/vits/wavfile.py,sha256=oQZiTIrdw0oLTbcVwKfGXye1WtKte6qK_52qVwiMvfc,26396
 phoonnx_train/vits/monotonic_align/__init__.py,sha256=5IdAOD1Z7UloMb6d_9NRFsXoNIjEQ3h9mvOSh_AtO3k,636
 phoonnx_train/vits/monotonic_align/setup.py,sha256=0K5iJJ2mKIklx6ncEfCQS34skm5hHPiz9vRlQEvevvY,266
-phoonnx-0.0.1a1.dist-info/METADATA,sha256=7ehavgd7HvRVp-U_k7Wqxewd58bwuriswuM0bDAh9bU,7868
-phoonnx-0.0.1a1.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
-phoonnx-0.0.1a1.dist-info/top_level.txt,sha256=ZrnHXe-4HqbOSX6fbdY-JiP7YEu2Bok9T0ji351MrmM,22
-phoonnx-0.0.1a1.dist-info/RECORD,,
+phoonnx-0.0.2a2.dist-info/METADATA,sha256=-uMrs5iV1A4gqenGg1r8ZyhZN1DhBd9HCsezIy4UxPE,8145
+phoonnx-0.0.2a2.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
+phoonnx-0.0.2a2.dist-info/top_level.txt,sha256=ZrnHXe-4HqbOSX6fbdY-JiP7YEu2Bok9T0ji351MrmM,22
+phoonnx-0.0.2a2.dist-info/RECORD,,

phoonnx_train/vits/lightning.py

@@ -299,9 +299,9 @@ class VitsModel(pl.LightningModule):
             test_audio = test_audio * (1.0 / max(0.01, abs(test_audio.max())))
 
             tag = test_utt.text or str(utt_idx)
-            self.logger.experiment.add_audio(
-                tag, test_audio, sample_rate=self.hparams.sample_rate
-            )
+           # self.logger.experiment.add_audio(
+           #     tag, test_audio, sample_rate=self.hparams.sample_rate
+           # )
 
         return val_loss