Create idn_phonemes.py

idn_phonemes.py

@@ -0,0 +1,92 @@
+import re
+
+ipa_map = {
+    "ng": "ŋ",
+    "ny": "ɲ",
+    "sy": "ʃ",
+    "kh": "x",
+    "c": "tʃ",
+    "j": "dʒ",
+    "y": "j",
+    "r": "r",
+    "x": "ks",
+    "a": "a",
+    "i": "i",
+    "u": "u",
+    "e": "ə",
+    "o": "o",
+    "b": "b",
+    "d": "d̪",
+    "t": "t̪",
+    "g": "ɡ",
+    "k": "k",
+    "p": "p",
+    "m": "m",
+    "n": "n",
+    "l": "l",
+    "s": "s",
+    "h": "h",
+    "w": "w",
+}
+
+num_words = {
+    0: "nol",
+    1: "satu",
+    2: "dua",
+    3: "tiga",
+    4: "empat",
+    5: "lima",
+    6: "enam",
+    7: "tujuh",
+    8: "delapan",
+    9: "sembilan",
+    10: "sepuluh",
+    11: "sebelas"
+}
+
+def number_to_words(n: int) -> str:
+    """Konversi angka 0–9999 ke kata dalam bahasa Indonesia"""
+    if n < 12:
+        return num_words[n]
+    elif n < 20:
+        return number_to_words(n-10) + " belas"
+    elif n < 100:
+        puluhan, sisa = divmod(n, 10)
+        result = number_to_words(puluhan) + " puluh"
+        if sisa:
+            result += " " + number_to_words(sisa)
+        return result
+    elif n < 200:
+        return "seratus" + (" " + number_to_words(n-100) if n > 100 else "")
+    elif n < 1000:
+        ratusan, sisa = divmod(n, 100)
+        result = number_to_words(ratusan) + " ratus"
+        if sisa:
+            result += " " + number_to_words(sisa)
+        return result
+    elif n < 2000:
+        return "seribu" + (" " + number_to_words(n-1000) if n > 1000 else "")
+    elif n < 10000:
+        ribuan, sisa = divmod(n, 1000)
+        result = number_to_words(ribuan) + " ribu"
+        if sisa:
+            result += " " + number_to_words(sisa)
+        return result
+    else:
+        return str(n)  # fallback
+
+def indo_to_ipa(text: str) -> str:
+    text = text.lower()
+    
+    # Cari semua angka dalam teks dan ubah ke kata
+    def replace_number(match):
+        num = int(match.group())
+        return number_to_words(num)
+    
+    text = re.sub(r"\d+", replace_number, text)
+    
+    # Konversi huruf → IPA
+    for k in sorted(ipa_map.keys(), key=lambda x: -len(x)):
+        text = re.sub(k, ipa_map[k], text)
+    
+    return "[" + text + "]"