A gem for inferior poetry
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



RubyRhymes is a gem that makes auto-generating terrible poetry a breeze. You use it roughly like so:

>> "RubyGems... is a gem quite sublime!".to_phrase.flat_rhymes
=> ["anticrime", "beim", "chime", "climb", "clime", "crime", ...]


Tested with Ruby 1.8.7


gem install ruby_rhymes


The easiest thing to do is embrace the String abuse:

>> "RubyRhymes is a gem quite sublime".to_phrase.flat_rhymes
=> ["anticrime", "beim", "chime", "climb", "clime", "crime", ...]

The first time you call most methods on a Phrase class (which is what these things are), the gem constructs the (static) in-memory dictionary, so expect a minor delay and some inevitable om-nom-nom-ing of memory.

Alright, what else we got? how about a some good ol' syllable countin', like in the old country?

>> "RubyRhymes is a gem quite sublime".to_phrase.syllables
=> 9

oh, how nice. Notice how even though RubyRhymes is not a dictionary word, we still got a pretty solid syllable estimate. That's what happens when you plagiarize from enough smart people on the internet. Note however that that's the extent of magic - calling .flat_rhymes on a non-dictionary word will return a lonely, desolate array:

>> "blog".to_phrase.flat_rhymes
=> []

Don't want to deal with silly phrases that don't end in dictionary words? there's a method for that:

>> "can i haz lolcats?".to_phrase.dict?
=> false

All by yourself one night, you may find that you're trying to figure out which sentences rhyme in a corpus of text:

>> "do you see a dog?".to_phrase.rhyme_keys
=> ["&C"]
>> "through the sailing fog!".to_phrase.rhyme_keys
=> ["&C"]

that's useful, you can now group_by (Rails) or whatever you're into. Notice I called .rhyme_keys instead of .rhyme_key. That's because words with multiple pronunciations may have multiple rhyme keys (tomato returns ["\2R", "\2:"]). You can still call .rhyme_key and that will just call .first for you - it's on the house. You'll get a nil/[] from these key/keys methods iff .dict? is false, so keep that in mind and don't assume that two words with a key of nil rhyme.

If you're not afraid of multiple pronunciations, you can call .rhymes instead of .flat_rhymes which will yield a map from rhyme key to rhymes. For a dictionary word with no rhymes (orange) it will still return a map to an empty array, whereas for .dict? == false expect {}

>> "RubyRhymes is a gem quite sublime".to_phrase.rhymes
=> {"+I"=>["anticrime", "beim", "chime", "climb", "clime", "crime", ...]}


  • Thomas, the co-author extraordinaire and Andre, my muse
  • The dictionaries, which is really what matters, came from Brian Langenberger, whose work was based on a CMU Pronouncing Dictionary.
  • The syllable counter for words that aren't in the dictionary comes from the source made available by Russell McVeigh. I just ported the PHP.

Modus Operandi

In essence we have three dictionaries available

  • multiple.txt : Word to pronunciation-encoding (ex. TOMATO => TOMATO TOMATO(2))
  • words.txt : Pronunciation-encoding to syllable rhyme key and syllable count (ex. TOMATO => \2R 3)
  • rhymes.txt : Rhyme-key to pronunciation-encodings of words that rhyme with it (ex. \2R => POTATO SAITO TOMATO)

from here on, it's fairly intuitive what needs to be done. multiple.txt is only has keys when the word actually has multiple pronunciations, so CAT will yield nothing there, meaning we can continue to check in words.txt.