-
Notifications
You must be signed in to change notification settings - Fork 1
/
msd_datamaker.py
168 lines (124 loc) · 7.26 KB
/
msd_datamaker.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
import os
import sys
import pickle
import numpy as np
import csv
import random
singer_data_dir = 'data/msd_data'
artist_id_to_name = pickle.load(open(os.path.join(singer_data_dir, 'artist_id_to_name.pkl'), 'rb'))
train_artists = [ artist_id.strip('\n') for artist_id in open(os.path.join(singer_data_dir, 'train_artists.txt'), 'r').readlines()]
random.shuffle(train_artists)
print (len(train_artists))
artist_to_train_songs = pickle.load(open(os.path.join(singer_data_dir, 'msd_artist_train_SVD.pkl'), 'rb'))
artist_to_valid_songs = pickle.load(open(os.path.join(singer_data_dir, 'msd_artist_valid_SVD.pkl'), 'rb'))
artist_to_test_songs = pickle.load(open(os.path.join(singer_data_dir, 'msd_artist_test_SVD.pkl'), 'rb'))
num_train_tracks = 15
num_valid_tracks = 18
# select which artist to train and eval with (unseen)
num_test_artist = 500 # num of unseen artists for all evaluation
num_train_artist = 1000 # num of training artists
version = '_a' # indicate data version
test_artists = train_artists[:num_test_artist] # change number
print (len(test_artists))
print (len(train_artists))
train_artists = train_artists[num_test_artist:num_test_artist+num_train_artist]
test_artist_to_int = {}
test_artist_to_name = {}
for a in range(len(test_artists)):
# print (artist_id_to_name[test_artists[a]]) # convert artist id to name
test_artist_to_int[test_artists[a]] = a
test_artist_to_name[test_artists[a]] = artist_id_to_name[test_artists[a]]
train_artist_to_int = {}
train_artist_to_name = {}
for a in range(len(train_artists):
train_artist_to_int[train_artists[a]] = a
train_artist_to_name[train_artists[a]] = artist_id_to_name[train_artists[a]]
# csv files
data_path = 'data'
fieldnames = ['artist_index', 'artist_id', 'artist_name', 'track_id', 'vocal_segments']
train_f = open(data_path + 'msd_train_data_' + str(num_train_singers) + version + '.csv', mode='w')
valid_f = open(data_path + 'msd_valid_data_' + str(num_train_singers) + version + '.csv', mode='w')
test_f = open(data_path + 'msd_test_data_' + str(num_train_singers) + version + '.csv', mode='w')
unseen_train_f = open(data_path + 'msd_unseen_model_data_' + str(num_test_artist) + version + '.csv', mode='w')
unseen_test_f = open(data_path + 'msd_unseen_eval_data_' + str(num_test_artist) + version + '.csv', mode='w')
train_writer = csv.DictWriter(train_f, fieldnames=fieldnames)
train_writer.writeheader()
valid_writer = csv.DictWriter(valid_f, fieldnames=fieldnames)
valid_writer.writeheader()
# test_writer_1000 = csv.DictWriter(test_f_1000, fieldnames=fieldnames)
# test_writer_1000.writeheader()
unseen_train_writer = csv.DictWriter(unseen_train_f, fieldnames=fieldnames)
unseen_train_writer.writeheader()
unseen_test_writer = csv.DictWriter(unseen_test_f, fieldnames=fieldnames)
unseen_test_writer.writeheader()
# artist for training
for i in range(len(train_artists)):
curr_artist_id = train_artists[i]
artist_train_songs = list(artist_to_train_songs[curr_artist_id].keys())
artist_valid_songs = list(artist_to_valid_songs[curr_artist_id].keys())
artist_test_songs = list(artist_to_test_songs[curr_artist_id].keys())
all_artist_songs = []
all_artist_songs.extend(artist_train_songs)
all_artist_songs.extend(artist_valid_songs)
all_artist_songs.extend(artist_test_songs)
random.shuffle(all_artist_songs)
# num_train_tracks = random.randint(5, 18)
# num_valid_tracks = num_train_tracks + 2
artist_train_songs = all_artist_songs[:num_train_tracks]
artist_valid_songs = all_artist_songs[num_train_tracks : num_valid_tracks]
# artist_test_songs = all_artist_songs[num_valid_tracks:]
for j in range(len(artist_train_songs)):
curr_track = artist_train_songs[j]
segments = artist_to_songs[curr_artist_id][curr_track]
middle10 = segments[len(segments)//2 -5 : len(segments)//2 +5]
if (len(middle10) < 10):
print ("warning, less than 10 segments")
print ('artist_id:', curr_artist_id, 'track_id:', curr_track, 'middle10:', middle10)
train_writer.writerow({'artist_index': train_artist_to_int[curr_artist_id], 'artist_id': curr_artist_id, 'artist_name': train_artist_to_name[curr_artist_id], 'track_id': curr_track, 'vocal_segments': middle10})
for j in range(len(artist_valid_songs)):
curr_track = artist_valid_songs[j]
segments = artist_to_songs[curr_artist_id][curr_track]
middle10 = segments[len(segments)//2 -5 : len(segments)//2 +5]
if (len(middle10) < 10):
print ("warning, less than 10 segments")
print ('artist_id:', curr_artist_id, 'track_id:', curr_track, 'middle10:', middle10)
valid_writer.writerow({'artist_index': train_artist_to_int[curr_artist_id], 'artist_id': curr_artist_id, 'artist_name': train_artist_to_name[curr_artist_id], 'track_id': curr_track, 'vocal_segments': middle10})
'''
for j in range(len(artist_test_songs)):
curr_track = artist_test_songs[j]
segments = artist_to_songs[curr_artist_id][curr_track]
middle10 = segments[len(segments)//2 -5 : len(segments)//2 +5]
if (len(middle10) < 10):
print ("warning, less than 10 segments")
print ('artist_id:', curr_artist_id, 'track_id:', curr_track, 'middle10:', middle10)
test_writer_1000.writerow({'artist_index': train_artist_to_int[curr_artist_id], 'artist_id': curr_artist_id, 'artist_name': train_artist_to_name[curr_artist_id], 'track_id': curr_track, 'vocal_segments': middle10})
'''
# artist as unseen data
for i in range(len(test_artists)):
curr_artist_id = test_artists[i]
artist_train_songs = list(artist_to_train_songs[curr_artist_id].keys())
artist_valid_songs = list(artist_to_valid_songs[curr_artist_id].keys())
artist_test_songs = list(artist_to_test_songs[curr_artist_id].keys())
all_artist_songs = []
all_artist_songs.extend(artist_train_songs)
all_artist_songs.extend(artist_valid_songs)
all_artist_songs.extend(artist_test_songs)
random.shuffle(all_artist_songs)
artist_train_songs = all_artist_songs[:15]
artist_test_songs = all_artist_songs[15:]
for j in range(len(artist_train_songs)):
curr_track = artist_train_songs[j]
segments = artist_to_songs[curr_artist_id][curr_track]
middle10 = segments[len(segments)//2 -5 : len(segments)//2 +5]
if (len(middle10) < 10):
print ("warning, less than 10 segments")
print ('artist_id:', curr_artist_id, 'track_id:', curr_track, 'middle10:', middle10)
unseen_train_writer.writerow({'artist_index': test_artist_to_int[curr_artist_id], 'artist_id': curr_artist_id, 'artist_name': test_artist_to_name[curr_artist_id], 'track_id': curr_track, 'vocal_segments': middle10})
for j in range(len(artist_test_songs)):
curr_track = artist_test_songs[j]
segments = artist_to_songs[curr_artist_id][curr_track]
middle10 = segments[len(segments)//2 -5 : len(segments)//2 +5]
if (len(middle10) < 10):
print ("warning, less than 10 segments")
print ('artist_id:', curr_artist_id, 'track_id:', curr_track, 'middle10:', middle10)
unseen_test_writer.writerow({'artist_index': test_artist_to_int[curr_artist_id], 'artist_id': curr_artist_id, 'artist_name': test_artist_to_name[curr_artist_id], 'track_id': curr_track, 'vocal_segments': middle10})