def gen_song_tuples(root_path):
# Parameters
# root_path: string -- path to MillionSongSubset's data folder
#
# Returns: lists of tuples where each tuple is a song (database row)
filename_re = "^[A-Z]{7}[0-9,A-F]{11}\.(h5|analysis)$" # Example: TRBIJIA128F425F57D.h5
for loop_nr, song_rec in enumerate(iterate_songs.iterate_folder_songs_extracted(root_path, filename_re)):
timbre = song_rec.timbre
sections_start = song_rec.sections_start
sections_conf = song_rec.sections_conf
segments_start = song_rec.segments_start
song_end = song_rec.song_end
artist = song_rec.artist
title = song_rec.title
id = song_rec.id
try:
feature_vector = get_feature_vector(timbre,sections_start,sections_conf,segments_start,song_end)
if feature_vector is None:
print 'skipping'
continue
start, end = get_start_end(sections_start, sections_conf, song_end)
tup = (unicode_if_str(title, 'utf-8'), unicode_if_str(artist, 'utf-8'),
unicode_if_str(id, 'utf-8'), start, end, unicode_if_str(features_as_str(feature_vector), 'utf-8'))
yield tup
except Exception as e:
sys.stderr.write('exception for {0}, {1}: {2}'.format(title, id, str(e)))
def save_feature_database(root_path,csvpath):
# Create a database where each song is represented by a line in a CSV file
# as such: artist/title , echonest id , feature_values
#
# Parameters
# root_path: string -- path to MillionSongSubset's data folder
# csvpath: string -- path where the .csv database will be created
#
# Returns: none
filename_re = "^[A-Z]{7}[0-9,A-F]{11}\.h5$" # Example: TRBIJIA128F425F57D.h5
time_start = time()
for loop_nr, song_rec in enumerate(iterate_songs.iterate_folder_songs_extracted(root_path, filename_re)):
btchromas = song_rec.btchromas
feature_vector,key = create_feature_vector(btchromas,'majmin')
artist = song_rec.artist
title = song_rec.title
id = song_rec.id
print id
print artist,':',title
print feature_vector
print_estimated_info(feature_vector,8,key)
save_feature_vector(feature_vector,artist,title,id,csvpath)
if ( (loop_nr + 1) % 1000) == 0:
print "{0} songs read in {1:.1f} seconds" \
.format(loop_nr + 1, time() - time_start)
end_time = time() - time_start
print "Total: {0} songs read in {1:.1f} seconds".format(loop_nr + 1, end_time)
#if __name__ == '__main__':
# Main program
#root_path = '/Users/victoriadennis/Documents/databases/MillionSongSubset/data'
#csvpath = '/Users/victoriadennis/Documents/databases/MillionSongSubset/features/songs.csv'
#filename_re = 'TRBIJIA128F425F57D.h5'
#save_feature_database(root_path,csvpath)
