README_flamenco.txt

1. Collecting and cleaning data samples (splitraw.py) 

	- splits data from flamenco website canteytoque.com
	- collect_audio_features (uses pyechonest to access song track, 		
artist information, geolocation), 
	- collect_letra_features (currently in .ipynb file. uses sci-kit-		
learn and nltk to collect NLP features: (tempo, artist, length of 
	verses, noun phrases/chunks) (sci_kit_learn)  

2. Run machine learning algorithms like tf-idf and clustering using 
	relevant song features	
	- some ideas are to cluster by palo, by artist/recording location
	- year song was published to learn features and style of songs, 		
artists 

3. Using JavaScript and libraries like d3 to visualize and connect the song data. 
	- connect songs based on features: 
	- chico or jondo...the assumption is that cante chico is light-	hearted and 
cante jondo is serious or tragic, but the verses can 		
	change the mood, and perhaps even the tonality within a particular 	
	recording
	- flamenco friends: which artists composed songs together, lived in 	
the same city, town or province