Double Metaphone
A phonetic-encoding algorithm that converts a name into up to two code representations based on how it sounds rather than how it is spelled, so names like Mohammed, Muhammad, and Mohamad can still match despite different spellings.
Double Metaphone improves on earlier phonetic algorithms (like the original Metaphone) by producing a primary and an alternate code for ambiguous spellings, which matters most for names transliterated from Arabic, Cyrillic, or other non-Latin scripts where there is no single correct Latin rendering. It is one layer in a broader fuzzy name-matching pipeline, typically run alongside token-based matching and edit-distance measures, because no single technique catches every source of name variation — transliteration, name-order differences, dropped middle names, data-entry errors, and deliberate obfuscation each need a different technique to catch.
Phonetic matching trades precision for recall: it is very good at catching a name that sounds the same but is spelled differently, and it will also surface unrelated names that happen to share a phonetic code, which is why phonetic-code matches are typically weighted as one signal among several (alongside name tokens, dates of birth, and other identifiers) in a combined match score rather than treated as a match on their own.