RhymerFinder

RhymerFinder predicts the rhymes in a song based on preceding lyrics by using gensim's Word2Vec implementation.

• Blog Post
• Web App (RhymerFinder.com)

Installation

python setup.py install

Accuracy Metrics

python rhymerfinder/run.py

Scraping

The lyrics corpus is provided in data/corpus.pkl. However, if you would like to re-scrape the lyrics into a Mongo database run the following:

To scrape the song URLs for each artist (outputed into data/html/urls.txt):

python rhymerfinder/scrape_artists.py

To scrape the HTML source, parse the lyrics, and store them using MongoDB (HTML source outputed to data/html/[artist name]):

python rhymerfinder/scrape_songs.py

Rhyme Your Own Lyrics

You can also use RhymerFinder to rhyme your own lyrics at the iPython command line (replicating the behavior of rhymerfinder.com).

First, cd into the "rhymerfinder" subdirectory". Train a rhymerfinder object with the following:

import rhymer_finder
import cPickle as pickle

#Initialize rhymer_finder object
rhymer = rhymer_finder.rhymer_finder()

print "Loading Corpus..."
with open('./data/corpus.pkl','r') as f:
    corpus = pickle.load(f)

print "Loading W2V..."
with open('./data/w2v.pkl','r') as f:
    rhymer.load_w2v(pickle.load(f))

print "Creating Rhyming Dictionary..."
rhymer.process_corpus(corpus)

Now you can use the "find_rhyme" method like so: rhymer.find_rhyme(Lyrics, Word_to_Rhyme)

For example,

rhymer.find_rhyme('You will see him on tv any given sunday win the superbowl and drive off in a','sunday')
      Cos-sim     Rhymes
0      0.41    hyundai
1      0.39      andre
2      0.37   driveway
3      0.35        ave
...    ...        ...