Fix k phonemization issues

g2p_id/__init__.py

@@ -20,5 +20,5 @@
 from .onnx_utils import WrapInferenceSession
 from .text_processor import TextProcessor
 
-__version__ = "0.4.0"
+__version__ = "0.4.1"
 __all__ = ["G2p", "LSTM", "BERT", "WrapInferenceSession", "TextProcessor"]

g2p_id/g2p.py

@@ -222,14 +222,14 @@ def __call__(self, text: str) -> List[List[str]]:
                 if isinstance(self.model, BERT):
                     pron = self._rule_based_g2p(pron)
 
-            prons.append(pron.split())
+            if pron.endswith("ʔ"):
+                pron = pron[:-1] + "k"
 
-        if pron.endswith("ʔ"):
-            pron = pron[:-1] + "k"
+            consonants = "bdjklmnprstwɲ"
 
-        consonants = "bdjklmnprstwɲ"
+            for letter in consonants:
+                text = text.replace(f"ʔ{letter}", f"k{letter}")
 
-        for letter in consonants:
-            text = text.replace(f"ʔ{letter}", f"k{letter}")
+            prons.append(pron.split())
 
         return prons

setup.py

@@ -14,7 +14,7 @@
 if __name__ == "__main__":
     setup(
         name="g2p_id_py",
-        version="0.4.0",
+        version="0.4.1",
         description="Indonesian G2P.",
         long_description=long_description,
         long_description_content_type="text/markdown",

tests/test_g2p.py

@@ -40,6 +40,7 @@ def test_g2p(g2p):
         ["s", "ə", "l", "a", "m", "a", "t"],
         ["d", "a", "t", "a", "ŋ"],
     ]
+    assert g2p("kakak layak") == [["k", "a", "k", "a", "k"], ["l", "a", "j", "a", "k"]]
 
 
 def test_rule_based_g2p(g2p):