Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Branch: master
Fetching contributors…

Cannot retrieve contributors at this time

167 lines (137 sloc) 6.151 kB
""" Processes track data from the Million Song Database. Specifically, this
file contains functions that load the flat-file format of tracks for the
MSD. The format is one track per line, where each line is represented by 54
fields as described here:
http://labrosa.ee.columbia.edu/millionsong/pages/field-list
except that in the flat file format, the 'track id' field has been moved
from field 52 to the first field.
A track is represented as a dictionary.
"""
import sys
import pprint
def load_track(line):
""" Loads a track from a single line """
t = {}
f = line.split('\t')
if len(f) == 54:
t['track_id'] = f[0]
t['analysis_sample_rate'] = f[1]
t['artist_7digitalid'] = f[2]
t['artist_familiarity'] = float(f[3])
t['artist_hotttnesss'] = float(f[4])
t['artist_id'] = f[5]
t['artist_latitude'] = float(f[6])
t['artist_location'] = f[7]
t['artist_longitude'] = float(f[8])
t['artist_mbid'] = f[9]
tag_words = f[10].split(',')
tag_count = f[11].split(',')
mbtags = [ (w, int(c)) for w,c in zip(tag_words, tag_count) if len(w) > 0]
t['artist_mbtags'] = mbtags
t['artist_name'] = f[12]
t['artist_playmeid'] = int(f[13])
artist_terms = f[14].split(',')
artist_terms_freq = f[15].split(',')
artist_terms_weight = f[16].split(',')
t['artist_terms'] = [ (term, float(freq), float(weight)) \
for term ,freq, weight in zip(artist_terms, artist_terms_freq, artist_terms_weight) if len(term) > 0]
t['audio_md5'] = f[17]
bars_confidence = f[18].split(',')
bars_start = f[19].split(',')
t['bars'] = [ (float(start), float(conf)) \
for start, conf in zip(bars_start, bars_confidence) if len(start) > 0 ]
beats_confidence = f[20].split(',')
beats_start = f[21].split(',')
t['beats'] = [ (float(start), float(conf)) \
for start, conf in zip(beats_start, beats_confidence) if len(start) > 0 ]
t['danceability'] = float(f[22])
t['duration'] = float(f[23])
t['end_of_fade_in'] = float(f[24])
t['energy'] = float(f[25])
t['key'] = (int(f[26]), float(f[27]))
t['loudness'] = float(f[28])
t['mode'] = (int(f[29]), float(f[30]))
t['release'] = f[31]
t['release_7digitalid'] = f[32]
srid = f[32].zfill(10)
t['cover_art'] = 'http://cdn.7static.com/static/img/sleeveart/%s/%s/%s/%s_200.jpg' \
% (srid[0:2], srid[2:5], srid[5:8], srid)
sections_confidence = f[33].split(',')
sections_start = f[34].split(',')
t['sections'] = [ (float(start), float(conf)) \
for start, conf in zip(sections_start, sections_confidence) if len(start) > 0 ]
seg_confidence = f[35].split(',')
seg_loudness_max = f[36].split(',')
seg_loudness_max_time = f[37].split(',')
seg_loudness_max_start = f[38].split(',')
seg_pitches = f[39].split(',')
seg_start = f[40].split(',')[:-1]
seg_timbre = f[41].split(',')
PITCH_COUNT = 12
TIMBRE_COUNT = 12
t['segments'] = []
for i, sstart in enumerate(seg_start):
if len(sstart) > 0:
seg = {}
seg['start'] = float(sstart)
seg['confidence'] = float(seg_confidence[i])
seg['loudness_max'] = float(seg_loudness_max[i])
seg['loudness_max_time'] = float(seg_loudness_max_time[i])
seg['loudness_start'] = float(seg_loudness_max_start[i])
seg['pitch'] =[ float(p) for p in seg_pitches[i * PITCH_COUNT: i * PITCH_COUNT + PITCH_COUNT]]
seg['timbre'] =[ float(p) for p in seg_timbre[i * TIMBRE_COUNT: i * TIMBRE_COUNT + TIMBRE_COUNT]]
t['segments'].append(seg)
if i < len(seg_start) - 1:
seg['duration'] = float(seg_start[i + 1]) - seg['start']
else:
seg['duration'] = t['duration'] - seg['start']
t['similar_artists'] = [s for s in f[42].split(',') if len(s) > 0]
t['song_hotttnesss'] = float(f[43])
t['song_id'] = f[44]
t['start_of_fade_out'] = float(f[45])
tatums_confidence = f[46].split(',')
tatums_start = f[47].split(',')
t['tatums'] = [ (float(start), float(conf)) \
for start, conf in zip(tatums_start, tatums_confidence) if len(start) > 0 ]
t['tempo'] = float(f[48])
t['time_signature'] = (int(f[49]), float(f[50]))
t['title'] = f[51]
t['track_7digitalid'] = int(f[52])
t['preview'] = 'http://previews.7digital.com/clips/34/%d.clip.mp3' % (int(f[52]), )
t['year'] = int(f[53])
return t
else:
print 'mismatched fields, found', len(f), 'should have 54'
return None
def load_tracks(path):
""" Loads a list of track from a file """
tracks = []
file = open(path)
for which, line in enumerate(file):
track = load_track(line)
if track <> None:
track['path'] = path
track['line'] = which
tracks.append(track)
file.close()
return tracks
def process_tracks(path, func):
""" applies func(track) to each track found in path """
file = open(path)
for which, line in enumerate(file):
track = load_track(line)
if track <> None:
track['path'] = path
track['line'] = which
func(track)
file.close()
def dump(track):
""" Dumps some data from a track for debugging """
print track['line'], track['track_id'], track['artist_id'], len(track['artist_mbtags']), \
len(track['artist_terms'] ), len(track['bars']), len(track['beats']), track['title'], \
track['key'], track['mode'], len(track['segments'])
for seg in track['segments']:
print ' ', seg['start'], seg['duration'], track['duration']
print
if __name__ == '__main__':
process_tracks(sys.argv[1], dump)
Jump to Line
Something went wrong with that request. Please try again.