update

README.md

@@ -1,6 +1,6 @@
 ---
-title: Khmer G2p Ipa
-emoji: 📉
+title: Khmer Phonemizer
+emoji: 🇰🇭
 colorFrom: purple
 colorTo: gray
 sdk: gradio

app.py

@@ -1,45 +1,39 @@
 import gradio as gr
 import phonetisaurus
-from collections import namedtuple
-
-model = phonetisaurus.Phonetisaurus(model="./model.fst")
-
-def Phoneticize (args) :
-    
-    results = model.Phoneticize (
-        args["token"],
-        args["nbest"],
-        args["beam"],
-        args["thresh"],
-        args["write_fsts"],
-        args["accumulate"],
-        args["pmass"]
-    )
-
-    for result in results :
-        uniques = [model.FindOsym (u) for u in result.Uniques]
-        return "".join(uniques)
-
-    return ""
-
+from collections import namedtuple    
+import re
+import shlex
+import tempfile
+import subprocess
 
 def phonemizer(text: str):
     
     if not text:
         return ""
     
-    args = {
-        "token": text,
-        "word": "word",
-        "nbest": 1,
-        "thresh": 10.0,
-        "write_fsts": False,
-        "accumulate": False,
-        "pmass": 0.0,
-        "beam": 500
-    }
+    env = phonetisaurus.guess_environment()
+    model_path = "./model.fst"
+
+    with tempfile.NamedTemporaryFile(suffix=".txt", mode="w+") as temp_file:
+        print(text, file=temp_file)
+        
+        temp_file.seek(0)
+
+        phonetisaurus_cmd = [
+            "phonetisaurus-apply",
+            "--model",
+            shlex.quote(str(model_path)),
+            "--word_list",
+            shlex.quote(str(temp_file.name)),
+            "--nbest",
+            str(1),
+        ]
+
+        result_str = subprocess.check_output(phonetisaurus_cmd, env=env, universal_newlines=True)
+        phoneme = result_str.split('\t')[1].strip()
+        phoneme = re.sub(r'\s', "", phoneme)
+        return f"/{phoneme}/"
 
-    return f"/{Phoneticize(args)}/"
 
 iface = gr.Interface(
     title="Khmer Phonemizer",