training/ar_lexicon.json

{
  "_comment": "Palestinian Arabic pronunciation lexicon: surface word -> diacritized rewrite fed to espeak. Add entries whenever a word sounds wrong; applied by prepare_arabic.normalize_arabic in both training prep and inference.",
  "حركتلو": "حَرَكتلو",
  "حركتله": "حَرَكتله",
  "قيمتها": "قِيمِتْها",
  "قيمته": "قِيمْتُه",
  "سبعمية": "سَبْعَمِيِّه",
  "ستمية": "سِتْمِيِّه",
  "خمسمية": "خَمَسْمِيِّه",
  "اربعمية": "أَرْبَعْمِيِّه",
  "تلتمية": "تَلَتْمِيِّه",
  "ميتين": "مِيتين",
  "مبارح": "مْبارِح",
  "سبعة": "سَبْعَه",
  "هلق": "هَلَق",
  "هلأ": "هَلَأ",
  "هيك": "هِيك",
  "وين": "وِين",
  "تنين": "تْنين",
  "قلت": "قُلْت",
  "معلش": "مَعليش",
  "معليش": "مَعليش",
  "بتذكر": "بِتذَكر",
  "بعرفش": "بَعْرَفِش",
  "يخرب": "يِخْرِب",
  "بدك": "بِدك",
  "قديش": "قَديش",
  "بستناك": "بَستَناك",
  "زلمة": "زَلَمِه"
}