SoundX/X

The SoundX/X ActiveX control offers Soundex, Extended Soundex, and Metaphone algorithms that can convert words or names to codes that represent how the words "sound". This is useful in a database application where users need to find names they may not know how to spell.

Soundex is a letter plus digits, Extended Soundex is purely numeric and can result in faster database scans, and Metaphone is letters, which can result in a better match rate, but also takes up more computer storage space. The length of the codes (Soundex, Extended Soundex and Metaphone) is selectable through a property.

To generate the code, simply set the Word property with the word or name for which you want a coded entry. When you do this, SoundX generates the codes for all three of the algorithms. All you have to do is read the appropriate property: Soundex, ExtSoundex, or Metaphone.

Soundex is an algorithm developed and patented by Margaret Odell and Robert Russell in the early part of this century (U.S. Patent 1,261,167 (1918) and 1,435,663 (1922)). Dont worry about the patents, theyve long since expired.

Many articles have been written about it, but one of the best descriptions is by Donald Knuth in The Art of Computer Programming, Vol. 3.

Soundex converts a word or name to a code comprised of a letter followed by three digits. Some redundancy is taken out of the word (such as stripping vowels, consonant doubles, etc.). The first letter is preserved and the code is generated from what remains.

This is a minor change to the basic Soundex algorithm. In this algorithm, the first letter is treated the same as all of the remaining letters (i.e., if it is a vowel, it is stripped; if it is part of a consonant double, the second is stripped (as in LLAMA), etc.).

This code is purely numeric. This can result in faster database scans to select like-sounding words or names.

Metaphone originated from Lawrence Philips, an artificial intelligence specialist at NAC Reinsurance. There is a write-up on Metaphone in the December 1990 issue of Computer Language.

Metaphone does a better job than Soundex and Extended Soundex when it comes to representing like-sounding names. Its drawback is that Metaphone is comprised purely of letters. In other words, it takes up more storage and, in a large database, this can be costly.