phoonnx 0.2.7a1__tar.gz → 0.3.0a1__tar.gz

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.2.7a1
+Version: 0.3.0a1
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx/config.py

@@ -56,6 +56,7 @@ class PhonemeType(str, Enum):
     OPENPHONEMIZER = "openphonemizer" # en
     G2PEN = "g2pen" # en
 
+    TUGAPHONE = "tugaphone"  # pt
     G2PFA = "g2pfa"
     OPENJTALK = "openjtalk" # ja
     CUTLET = "cutlet" # ja
@@ -395,7 +396,7 @@ def get_phonemizer(phoneme_type: PhonemeType,
     from phoonnx.phonemizers import (EpitranPhonemizer, EspeakPhonemizer, OpenPhonemizer, OpenJTaklPhonemizer,
                                      ByT5Phonemizer, CharsiuPhonemizer, DeepPhonemizer, PersianPhonemizer,
                                      G2pCPhonemizer, G2pMPhonemizer, G2PKPhonemizer, G2PEnPhonemizer,
-                                     TransphonePhonemizer, MirandesePhonemizer, GoruutPhonemizer,
+                                     TransphonePhonemizer, MirandesePhonemizer, GoruutPhonemizer, TugaphonePhonemizer,
                                      GruutPhonemizer, GraphemePhonemizer, MantoqPhonemizer, MisakiPhonemizer,
                                      KoG2PPhonemizer, PypinyinPhonemizer, PyKakasiPhonemizer, CotoviaPhonemizer,
                                      CutletPhonemizer, PhonikudPhonemizer, VIPhonemePhonemizer, XpinyinPhonemizer,
@@ -404,6 +405,8 @@ def get_phonemizer(phoneme_type: PhonemeType,
         phonemizer = EspeakPhonemizer()
     elif phoneme_type == PhonemeType.BYT5:
         phonemizer = ByT5Phonemizer(model)
+    elif phoneme_type == PhonemeType.TUGAPHONE:
+        phonemizer = TugaphonePhonemizer()
     elif phoneme_type == PhonemeType.CHARSIU:
         phonemizer = CharsiuPhonemizer(model)
     elif phoneme_type == PhonemeType.GRUUT:

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx/phonemizers/__init__.py

@@ -7,6 +7,7 @@ from phoonnx.phonemizers.vi import VIPhonemePhonemizer
 from phoonnx.phonemizers.he import PhonikudPhonemizer
 from phoonnx.phonemizers.ar import MantoqPhonemizer
 from phoonnx.phonemizers.fa import PersianPhonemizer
+from phoonnx.phonemizers.pt import TugaphonePhonemizer
 from phoonnx.phonemizers.ja import PyKakasiPhonemizer, CutletPhonemizer, OpenJTaklPhonemizer
 from phoonnx.phonemizers.ko import KoG2PPhonemizer, G2PKPhonemizer
 from phoonnx.phonemizers.zh import (G2pCPhonemizer, G2pMPhonemizer, PypinyinPhonemizer,
@@ -18,6 +19,7 @@ from phoonnx.phonemizers.mwl import MirandesePhonemizer
 Phonemizer = Union[
     MisakiPhonemizer,
     ByT5Phonemizer,
+    TugaphonePhonemizer,
     UnicodeCodepointPhonemizer,
     CharsiuPhonemizer,
     EspeakPhonemizer,

phoonnx-0.3.0a1/phoonnx/phonemizers/pt.py

@@ -0,0 +1,57 @@
+from phoonnx.phonemizers.base import BasePhonemizer
+from phoonnx.config import Alphabet
+
+
+class TugaphonePhonemizer(BasePhonemizer):
+
+    def __init__(self):
+        from tugaphone import TugaPhonemizer
+        self.tuga = TugaPhonemizer()
+        super().__init__(Alphabet.IPA)
+
+    @classmethod
+    def get_lang(cls, target_lang: str) -> str:
+        """
+        Validates and returns the closest supported language code.
+
+        Args:
+            target_lang (str): The language code to validate.
+
+        Returns:
+            str: The validated language code.
+
+        Raises:
+            ValueError: If the language code is unsupported.
+        """
+        # this check is here only to throw an exception if invalid language is provided
+        return cls.match_lang(target_lang, ["pt-PT", "pt-BR", "pt-AO", "pt-MZ", "pt-TL"])
+
+    def phonemize_string(self, text: str, lang: str) -> str:
+        lang = self.get_lang(lang)
+        # TODO - support regional dialects
+        return self.tuga.phonemize_sentence(text, lang)
+
+
+
+if __name__ == "__main__":
+
+    pho = TugaphonePhonemizer()
+
+    sentences = [
+        "O gato dorme.",
+        "Tu falas português muito bem.",
+        "O comboio chegou à estação.",
+        "A menina comeu o pão todo.",
+        "Vou pôr a manteiga no frigorífico.",
+        "Ele está a trabalhar no escritório.",
+        "Choveu muito ontem à noite.",
+        "A rapariga comprou um telemóvel novo.",
+        "Vamos tomar um pequeno-almoço.",
+        "O carro ficou sem gasolina."
+    ]
+
+    for s in sentences:
+        print(s)
+        for code in ["pt-PT", "pt-BR", "pt-AO", "pt-MZ", "pt-TL"]:
+            print(f"{code} → {pho.phonemize_string(s, code)}")
+        print("######")

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx/version.py

@@ -1,7 +1,7 @@
 # START_VERSION_BLOCK
 VERSION_MAJOR = 0
-VERSION_MINOR = 2
-VERSION_BUILD = 7
+VERSION_MINOR = 3
+VERSION_BUILD = 0
 VERSION_ALPHA = 1
 # END_VERSION_BLOCK
 

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.2.7a1
+Version: 0.3.0a1
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx.egg-info/SOURCES.txt

@@ -26,6 +26,7 @@ phoonnx/phonemizers/ja.py
 phoonnx/phonemizers/ko.py
 phoonnx/phonemizers/mul.py
 phoonnx/phonemizers/mwl.py
+phoonnx/phonemizers/pt.py
 phoonnx/phonemizers/vi.py
 phoonnx/phonemizers/zh.py
 phoonnx/thirdparty/__init__.py
@@ -93,6 +94,7 @@ phoonnx_train/../phoonnx/phonemizers/ja.py
 phoonnx_train/../phoonnx/phonemizers/ko.py
 phoonnx_train/../phoonnx/phonemizers/mul.py
 phoonnx_train/../phoonnx/phonemizers/mwl.py
+phoonnx_train/../phoonnx/phonemizers/pt.py
 phoonnx_train/../phoonnx/phonemizers/vi.py
 phoonnx_train/../phoonnx/phonemizers/zh.py
 phoonnx_train/../phoonnx/thirdparty/__init__.py

{phoonnx-0.2.7a1 → phoonnx-0.3.0a1}/phoonnx.egg-info/requires.txt

@@ -236,6 +236,7 @@ epitran
 [pt]
 gruut[pt]<3.0,>=2.3.0
 epitran
+tugaphone
 
 [quy]
 epitran