ABSTRACT
After hours of staring fruitlessly at a blank sheet of paper, waiting patiently for inspiration to strike, I thought to myself, “Rap is very hard to write.”
Luckily for rappers-on-the-rise such as myself, Hieu Nguyen and Brian Sa of Stanford took a corpus of rap lyrics and wrote a rap lyric generator. The generator can produce lines on a specific theme incorporating rhyme and structure.
RESULTS
Surprisingly enough, the generator’s output often (though not always) passes an intuitive Turing test.
there’s nothing like a pretty hoe on her knees
suckin’ my d yeaaah. and lickin’ my b’s
we don’t have to take our clothes off to bust a nut
when i pull out my dick biiiitch pucker up
But the generator’s output is not perfect.
she got on her knees and gave some good hot head yeah hot head
hoes some white some niggeroes
but i like the ones who suck toes and assholes
with tongues like razors that cut when she licks ooh
Although the generator did simulate ethnic diversity among head-givers (a crucial point), the last line of output indicates that the generator may not understand what constitutes “good hot head.”
At the end of the day, however, the computer demonstrates a keen understanding of the male psyche:
i like getting head ’cause it’s so convenient huh
you can do it any time you don’t have to beat it
True enough, computer. True enough.
[note from Kayleigh Roberts about that last line: “technically, one CAN get head anytime, but likely will not anytime one wants it. again, I speak as an outside observer, but I infer that there are times it is wanted and not received.”]
NOTES
Note #1: The generator actually generates numerous candidate lines, then picks a winning line according to a score. The criteria for scoring:
1. The log probability of the sentence from our language model, divided by sentence length
2. The log probability of the sentence length
3. The sum of logs of TFICF (term frequency-inverse corpus frequency) of each word in the sentence
4. Whether the last word of the line rhymed with the last word of the previous line
5. Whether the last word of the line rhymed with another word in the sentence
6. Whether the last word of the line had the same number of syllables as the last word of the previous line
Note #2: My personal favorite line among generated text:
cause i eat up tracks like hannibal and dahmer
CONCLUSION
This generator allows us to construct a list of rappers who are worse at rapping than a computer.
And so on.
[read the original paper here.]
