phoonnx 0.1.0a3__tar.gz → 0.1.1a1__tar.gz

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.1.0a3
+Version: 0.1.1a1
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com
@@ -41,6 +41,7 @@ Provides-Extra: jam
 Provides-Extra: hsn
 Provides-Extra: fil
 Provides-Extra: ml
+Provides-Extra: mwl
 Provides-Extra: lez
 Provides-Extra: si
 Provides-Extra: kk

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/README.md

@@ -89,6 +89,7 @@ wrap, and the output alphabets they can generate.
 | **Multilingual**    | `MisakiPhonemizer`     | [misaki](https://github.com/hexgrad/misaki)                                                                        | IPA                          | Misaki is a G2P engine designed for Kokoro models.                                                                                          |
 | **Multilingual**    | `TransphonePhonemizer` | [transphone](https://github.com/xinjli/transphone)                                                                 | IPA                          | It provides approximated phoneme tokenizers and G2P model for 7546 languages registered in the Glottolog database.                          |
 | **Multilingual**    | `EpitranPhonemizer`    | [epitran](https://github.com/dmort27/epitran)                                                                      | IPA                          | A tool for transcribing orthographic text as IPA                                                                                            |
+| **Mirandese (mwl)** | `MirandesePhonemizer`  | [mwl_phonemizer](https://github.com/TigreGotico/mwl_phonemizer)                                                                 | IPA                          | A tool for transcribing orthographic text as IPA                                                                                            |
 | **Arabic (ar)**     | `MantoqPhonemizer`     | [mantoq](https://github.com/mush42/mantoq)                                                                         | BUCKWALTER, IPA              | Translates unvoweled Arabic to phonemes, with optional conversion to IPA.                                                                   |
 | **Chinese (zh)**    | `JiebaPhonemizer`      | [jieba](https://github.com/fxsjy/jieba)                                                                            | HANZI                        | Segments Chinese text into words with spaces; useful for pre-processing.                                                                    |
 | **Chinese (zh)**    | `G2pMPhonemizer`       | [g2pC](https://github.com/Kyubyong/g2pC)                                                                           | IPA, Pinyin                  | CRF-based Grapheme-to-Phoneme converter                                                                                                     |

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx/config.py

@@ -48,6 +48,8 @@ class PhonemeType(str, Enum):
     EPITRAN = "epitran"
     BYT5 = "byt5"
     CHARSIU = "charsiu"  # technically same as byt5, but needs special handling for whitespace
+    TRANSPHONE = "transphone"
+    MIRANDESE = "mwl_phonemizer"
 
     DEEPPHONEMIZER = "deepphonemizer" # en
     OPENPHONEMIZER = "openphonemizer" # en
@@ -392,6 +394,7 @@ def get_phonemizer(phoneme_type: PhonemeType,
     from phoonnx.phonemizers import (EpitranPhonemizer, EspeakPhonemizer, OpenPhonemizer, OpenJTaklPhonemizer,
                                      ByT5Phonemizer, CharsiuPhonemizer, DeepPhonemizer, PersianPhonemizer,
                                      G2pCPhonemizer, G2pMPhonemizer, G2PKPhonemizer, G2PEnPhonemizer,
+                                     TransphonePhonemizer, MirandesePhonemizer,
                                      GruutPhonemizer, GraphemePhonemizer, MantoqPhonemizer, MisakiPhonemizer,
                                      KoG2PPhonemizer, PypinyinPhonemizer, PyKakasiPhonemizer, CotoviaPhonemizer,
                                      CutletPhonemizer, PhonikudPhonemizer, VIPhonemePhonemizer, XpinyinPhonemizer,
@@ -408,6 +411,10 @@ def get_phonemizer(phoneme_type: PhonemeType,
         phonemizer = EpitranPhonemizer()
     elif phoneme_type == PhonemeType.MISAKI:
         phonemizer = MisakiPhonemizer()
+    elif phoneme_type == PhonemeType.TRANSPHONE:
+        phonemizer = TransphonePhonemizer()
+    elif phoneme_type == PhonemeType.MIRANDESE:
+        phonemizer = MirandesePhonemizer()
     elif phoneme_type == PhonemeType.DEEPPHONEMIZER:
         phonemizer = DeepPhonemizer(model)
     elif phoneme_type == PhonemeType.OPENPHONEMIZER:

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx/phonemizers/__init__.py

@@ -12,7 +12,8 @@ from phoonnx.phonemizers.ko import KoG2PPhonemizer, G2PKPhonemizer
 from phoonnx.phonemizers.zh import (G2pCPhonemizer, G2pMPhonemizer, PypinyinPhonemizer,
                                     XpinyinPhonemizer, JiebaPhonemizer)
 from phoonnx.phonemizers.mul import (EspeakPhonemizer, EpitranPhonemizer, MisakiPhonemizer,
-                                     GruutPhonemizer, ByT5Phonemizer, CharsiuPhonemizer)
+                                     GruutPhonemizer, ByT5Phonemizer, CharsiuPhonemizer, TransphonePhonemizer)
+from phoonnx.phonemizers.mwl import MirandesePhonemizer
 
 Phonemizer = Union[
     MisakiPhonemizer,
@@ -22,6 +23,8 @@ Phonemizer = Union[
     EspeakPhonemizer,
     GruutPhonemizer,
     EpitranPhonemizer,
+    TransphonePhonemizer,
+    MirandesePhonemizer,
     OpenJTaklPhonemizer,
     CutletPhonemizer,
     PyKakasiPhonemizer,

phoonnx-0.1.1a1/phoonnx/phonemizers/mwl.py

@@ -0,0 +1,36 @@
+from phoonnx.phonemizers.base import BasePhonemizer, Alphabet
+from mwl_phonemizer import CRFOrthoCorrector
+
+
+class MirandesePhonemizer(BasePhonemizer):
+    _LANGS = ["mwl"]
+
+    def __init__(self):
+        super().__init__(Alphabet.IPA)
+        self.pho = CRFOrthoCorrector()
+
+    @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.
+        """
+        return cls.match_lang(target_lang, cls._LANGS)
+
+    def phonemize_string(self, text: str, lang: str) -> str:
+        # Validate language is supported
+        lang = self.get_lang(lang)
+        return self.pho.phonemize_sentence(text)
+
+
+if __name__ == "__main__":
+    pho = MirandesePhonemizer()
+    print(pho.phonemize_string("ls", "mwl"))

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx/version.py

@@ -1,10 +1,10 @@
 # START_VERSION_BLOCK
 VERSION_MAJOR = 0
 VERSION_MINOR = 1
-VERSION_BUILD = 0
-VERSION_ALPHA = 3
+VERSION_BUILD = 1
+VERSION_ALPHA = 1
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"
 if VERSION_ALPHA:
-    VERSION_STR += f"a{VERSION_ALPHA}"
+    VERSION_STR += f"a{VERSION_ALPHA}"

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.1.0a3
+Version: 0.1.1a1
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com
@@ -41,6 +41,7 @@ Provides-Extra: jam
 Provides-Extra: hsn
 Provides-Extra: fil
 Provides-Extra: ml
+Provides-Extra: mwl
 Provides-Extra: lez
 Provides-Extra: si
 Provides-Extra: kk

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx.egg-info/SOURCES.txt

@@ -25,6 +25,7 @@ phoonnx/phonemizers/he.py
 phoonnx/phonemizers/ja.py
 phoonnx/phonemizers/ko.py
 phoonnx/phonemizers/mul.py
+phoonnx/phonemizers/mwl.py
 phoonnx/phonemizers/vi.py
 phoonnx/phonemizers/zh.py
 phoonnx/thirdparty/__init__.py
@@ -91,6 +92,7 @@ phoonnx_train/../phoonnx/phonemizers/he.py
 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/vi.py
 phoonnx_train/../phoonnx/phonemizers/zh.py
 phoonnx_train/../phoonnx/thirdparty/__init__.py

{phoonnx-0.1.0a3 → phoonnx-0.1.1a1}/phoonnx.egg-info/requires.txt

@@ -206,6 +206,9 @@ epitran
 [mt]
 epitran
 
+[mwl]
+mwl_phonemizer
+
 [my]
 epitran