Arabic Text Toolkit: Normalize, Transliterate, Numbers
Pricing
from $2.00 / 1,000 results
Normalise Arabic text, strip diacritics, transliterate Arabic to Latin, match names across scripts, and write amounts in Arabic words. Pure computation, no web calls.
Texts to process, one per line (up to 10,000, each up to 20,000 characters). Each text is one billed result.
What to compute for every text: detect (script and diacritics report), normalize, transliterate (Arabic to Latin), name_key (fuzzy match key for Arabic or Latin names).
[ "detect", "normalize", "transliterate"]Normalize: strip tashkeel (fatha, damma, kasra, shadda, sukun, tanween, dagger alef).
Normalize: remove the elongation character (ـ).
Normalize: map أ إ آ ٱ to plain ا (useful for search matching).
Normalize: map ى to ي (useful for search matching).
Normalize: map ة to ه (aggressive; useful for search matching only).
Normalize: convert digits to Western (0-9), Arabic-Indic (٠-٩) or leave unchanged. Persian digits are read as input too.
Transliterate: use chat digits for letters without a Latin equivalent (ع=3, ح=7, خ=5, ء=2, ط=6).
Numbers or amounts such as 1250.50 or ١٢٣٤٫٥٠ (up to 10,000). Each amount is one billed result. Max 999,999,999,999.