-
Notifications
You must be signed in to change notification settings - Fork 7
/
create_data_pickle.py
192 lines (145 loc) · 6.6 KB
/
create_data_pickle.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
from common import load_track, GENRES, NUMTRACKS, GENRE_IDS, GENRE_IDS_STRS
import sys
import numpy as np
from math import pi
#import _pickle as cPickle
from cPickle import dump
import os
import pandas as pd
from optparse import OptionParser
import librosa
from progress.bar import Bar
TRACK_COUNT = 1000
MEL_KWARGS = {
'n_fft': 2048,
'hop_length': 2048//2,
#'n_mels': 128
}
def get_default_shape(dataset_path):
tmp_features, _ = load_track(os.path.join(dataset_path,
'000002.mp3'))
return tmp_features.shape
def collect_data(dataset_path, genre_num):
'''
Collects data from the GTZAN dataset into a pickle. Computes a Mel-scaled
power spectrogram for each track.
:param dataset_path: path to the GTZAN dataset directory
:returns: triple (x, y, track_paths) where x is a matrix containing
extracted features, y is a one-hot matrix of genre labels and
track_paths is a dict of absolute track paths indexed by row indices in
the x and y matrices
'''
print genre_num
print [GENRE_IDS_STRS[genre_num]]
tracks = pd.read_csv('data/track_to_genre.csv');
tracks = tracks[np.isfinite(tracks['genre_id'])]
print len(tracks)
reduced_tracks = tracks.loc[tracks['genre_id'] == GENRE_IDS[genre_num]]
print len(reduced_tracks)
num_tracks = len(reduced_tracks)
default_shape = get_default_shape(dataset_path)
print "default:",default_shape
x = np.zeros((num_tracks,) + default_shape, dtype=np.float32)
y = np.zeros((num_tracks, len(GENRE_IDS)), dtype=np.float32)
track_paths = {}
#print(tracks)
#print(NUMTRACKS[0])
reduced_tracks = reduced_tracks.sort_values(by=['genre_id'])
print reduced_tracks
curr_genre = None
song_ID = ""
i = 0
bar = Bar('Progress:', max=len(reduced_tracks))
for row in reduced_tracks.iterrows():
#print "Processing Row : " , row
genre = int(row[1]['genre_id'])
song_ID = str(int(row[1][1]))
if genre != curr_genre:
curr_genre = genre
z = str(song_ID).zfill(6)
file_name = z+".mp3"
#file_name = "132567.mp3"
path = os.path.join(dataset_path, file_name)
if os.path.isfile(path):
track_index = i#curr_genre * (num_tracks // len(GENRES)) + i
genre_index = GENRE_IDS.index(curr_genre)
#print curr_genre, "genre_index:", genre_index
try:
d, sample_rate = librosa.load(path, sr=None, mono=True)#load_track(path, default_shape)
#stft = np.abs(librosa.stft(x, n_fft=2048, hop_length=512))
#mel = librosa.feature.melspectrogram(sr=sample_rate, S=stft**2).T
#log_mel = librosa.logamplitude(mel)
features = librosa.feature.melspectrogram(d, **MEL_KWARGS).T
if features.shape[0] < default_shape[0]:
delta_shape = (default_shape[0] - features.shape[0],
default_shape[1])
features = np.append(features, np.zeros(delta_shape), axis=0)
elif features.shape[0] > default_shape[0]:
features = features[: default_shape[0], :]
features[features == 0] = 1e-6
features = np.log(features)
x[track_index] = features
y[track_index, genre_index] = 1
#print y[track_index]
track_paths[track_index] = os.path.abspath(path)
i += 1
except Exception, e:
print "Corrupted File at" + path
print e
bar.next()
bar.finish()
return (x, y, track_paths)
'''
file_name = z+".mp3"
print('Processing', file_name)
path = os.path.join(dataset_path, file_name)
track_index = genre_index * (num_tracks // len(GENRES)) + i
#x[track_index], _ = librosa.load(path, sr=None, mono=False)#load_track(path, default_shape)
x, sample_rate = librosa.load(path, sr=None, mono=True)#load_track(path, default_shape)
#print(x[track_index])
#print(x[1000:20000])
stft = np.abs(librosa.stft(x, n_fft=2048, hop_length=512))
mel = librosa.feature.melspectrogram(sr=sample_rate, S=stft**2)
log_mel = librosa.logamplitude(mel)
#librosa.display.specshow(log_mel, sr=sample_rate, hop_length=512, x_axis='time', y_axis='mel')
print(log_mel)
y[track_index, genre_index] = 1
track_paths[track_index] = os.path.abspath(path)
#Process
for (genre_index, genre_name) in enumerate(GENRES):
#temp_
for i in xrange(NUMTRACKS[genre_index]):
z = str(i).zfill(6)
file_name = z+'.mp3' #need to fix filename to our format.
print('Processing', file_name)
path = os.path.join(dataset_path, file_name)
track_index = genre_index * (num_tracks // len(GENRES)) + i
#x[track_index], _ = librosa.load(path, sr=None, mono=False)#load_track(path, default_shape)
x, sample_rate = librosa.load(path, sr=None, mono=True)#load_track(path, default_shape)
#print(x[track_index])
#print(x[1000:20000])
stft = np.abs(librosa.stft(x, n_fft=2048, hop_length=512))
mel = librosa.feature.melspectrogram(sr=sample_rate, S=stft**2)
log_mel = librosa.logamplitude(mel)
#librosa.display.specshow(log_mel, sr=sample_rate, hop_length=512, x_axis='time', y_axis='mel')
print(log_mel)
y[track_index, genre_index] = 1
track_paths[track_index] = os.path.abspath(path)
'''
if __name__ == '__main__':
parser = OptionParser()
#Change default = ps.path.join(os.path.dirname(__file___, 'data/genres')
parser.add_option('-d', '--dataset_path', dest='dataset_path',
default=os.path.join(os.path.dirname(__file__), '/Users/Andrew/Downloads/fma'),
help='path to the GTZAN dataset directory', metavar='DATASET_PATH')
parser.add_option('-o', '--output_pkl_path', dest='output_pkl_path',
default=os.path.join(os.path.dirname(__file__), 'data/data.pkl'), #/Volumes/G-DRIVE mob/pickled_spect/data.pkl
help='path to the output pickle', metavar='OUTPUT_PKL_PATH')
options, args = parser.parse_args()
(x, y, track_paths) = collect_data(options.dataset_path, int(sys.argv[1]))
data = {'x': x, 'y': y, 'track_paths': track_paths}
#print(data)
print(data['x'])
with open(options.output_pkl_path, 'w') as f:
dump(data, f)
f.close()