Ya, could someone explain how this algorithm works?

There are rhyming dictionaries, but also just parsing the pronunciation indicators of a traditional dictionary for matching end-sounds would work -- and allow matching syllable counts as well. Wiktionary includes pronunciation guides in 'IPA' symbols:

http://en.wikipedia.org/wiki/International_Phonetic_Alphabet

There's also 'soundex' encoding of names/words...

http://en.wikipedia.org/wiki/Soundex

Ahh... I see. I didn't even know such dictionaries existed! Thanks!

I assume it just looks it up in a rhyming dictionary.