Soundex is not technically a Cipher or Code. It’s invention was for census data.
The idea behind Soundex (or SNDX, although not usually referenced in that manner) is to provide a representation of a name that will match for any minor variations that do not greatly alter the pronounciation. Pronunciation. Whichever.
The idea being that Robert and Rupert will be represented the same (R163) but Rubin will differ (R150). Because it was intended for names, it’s function is limited outside that scope. The original soundex structure calls for first letter followed by 3 digits, making the algorithm questionable for large words.
- Keep first letter of word or name, this is the first value in the Soundex value
- For remaining letters, convert all consonants with the table after the jump.
- All adjacent duplicate digits are collapsed into a single instance of each
- All remaining characters (vowels) are dropped
- The first three numbers are the remainder of the Soundex Value