-
Notifications
You must be signed in to change notification settings - Fork 61
/
Copy pathutils.py
540 lines (442 loc) · 22.1 KB
/
utils.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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
import math
import midi
import pretty_midi
import json
import os
import numpy as np
import pandas as pd
from gensim.models import Word2Vec
def get_syll_data(full_data, path_model, songlength, num_songs):
syllModel = Word2Vec.load(path_model)
syll2Vec = syllModel.wv['yo']
full_data = full_data[:, 0]
syll_matrix = np.zeros(shape=(num_songs, len(syll2Vec) * songlength))
for i in range(0, num_songs):
for j in range(0, songlength):
try:
syll_matrix[i][j * len(syll2Vec):(j + 1) * len(syll2Vec)] = syllModel.wv[full_data[i][0][j][0]]
except:
a = 0
return syll_matrix
def get_midi_data(full_data, songlength, num_songs, num_midi_features):
full_data = full_data[:, 0]
data_matrix = np.zeros(shape=(num_songs, num_midi_features * songlength))
for i in range(0, num_songs):
for j in range(0, songlength):
for feat in range(0, num_midi_features):
try:
data_matrix[i][j * num_midi_features + feat] = full_data[i][0][j][1][feat]
except:
data_matrix[i][j * num_midi_features + feat] = 0
return data_matrix
def load_settings_from_file(settings):
"""
Handle loading settings from a JSON file, filling in missing settings from
the command line defaults, but otherwise overwriting them.
"""
settings_path = './settings/' + settings['settings_file'] + '.txt'
print('Loading settings from', settings_path)
settings_loaded = json.load(open(settings_path, 'r'))
# check for settings missing in file
for key in settings.keys():
if not key in settings_loaded:
print(key, 'not found in loaded settings - adopting value from command line defaults: ', settings[key])
# overwrite parsed/default settings with those read from file, allowing for
# (potentially new) default settings not present in file
settings.update(settings_loaded)
return settings
def cents_to_pitchwheel_units(cents):
return int(40.96*(float(cents)))
def tone_to_freq(tone):
"""
returns the frequency of a tone.
formulas from
* https://en.wikipedia.org/wiki/MIDI_Tuning_Standard
* https://en.wikipedia.org/wiki/Cent_(music)
"""
return math.pow(2, ((float(tone)-69.0)/12.0)) * 440.0
def freq_to_tone(freq):
"""
returns a dict d where
d['tone'] is the base tone in midi standard
d['cents'] is the cents to make the tone into the exact-ish frequency provided.
multiply this with 8192 to get the midi pitch level.
formulas from
* https://en.wikipedia.org/wiki/MIDI_Tuning_Standard
* https://en.wikipedia.org/wiki/Cent_(music)
"""
if freq <= 0.0:
return None
float_tone = (69.0+12*math.log(float(freq)/440.0, 2))
int_tone = int(float_tone)
cents = int(1200*math.log(float(freq)/tone_to_freq(int_tone), 2))
return {'tone': int_tone, 'cents': cents}
def save_midi_pattern(filename, midi_pattern):
if filename is not None:
midi.write_midifile(filename, midi_pattern)
def get_batch_multi(batch_size, pointer, train, validate, test, num_midi_features, num_syllables_per_sentence,
part='train'):
batch_songs = []
batch_meta = []
batch_meta_wrong = []
if part == 'train':
batch_size = min(batch_size, np.shape(train[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(train[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(train[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(train[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(train[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
elif part == 'validation':
batch_size = min(batch_size, np.shape(validate[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(validate[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(validate[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(validate[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(validate[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
else:
batch_size = min(batch_size, np.shape(test[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(test[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(test[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(test[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(test[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
try:
batch_songs_list = np.split(np.asarray(batch_songs), indices_or_sections=num_syllables_per_sentence, axis=1)
batch_songs_list = np.transpose(batch_songs_list, axes=(1, 0, 2))
batch_meta_list = np.split(np.asarray(batch_meta), indices_or_sections=num_syllables_per_sentence, axis=1)
batch_meta_list = np.transpose(batch_meta_list, axes=(1, 0, 2))
batch_meta_wrong_list = np.split(np.asarray(batch_meta_wrong), indices_or_sections=num_syllables_per_sentence,
axis=1)
batch_meta_wrong_list = np.transpose(batch_meta_wrong_list, axes=(1, 0, 2))
except:
batch_songs_list = None
batch_meta_list = None
batch_meta_wrong_list = None
pointer = pointer + batch_size
return batch_songs_list, batch_meta_list, batch_meta_wrong_list, pointer
def get_batch(batch_size, pointer, train, validate, test, num_midi_features, num_syllables_per_sentence,
num_syllables_features, part='train'):
batch_songs = []
batch_meta = []
batch_meta_wrong = []
if part == 'train':
batch_size = min(batch_size, np.shape(train[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(train[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(train[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(train[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(train[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
elif part == 'validation':
batch_size = min(batch_size, np.shape(validate[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(validate[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(validate[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(validate[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(validate[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
else:
batch_size = min(batch_size, np.shape(test[pointer:None])[0])
for i in range(pointer, pointer+batch_size):
try:
batch_songs.append(test[i][0:num_midi_features*num_syllables_per_sentence])
batch_meta.append(test[i][num_midi_features*num_syllables_per_sentence:None])
if i == pointer:
batch_meta_wrong.append(test[i+batch_size-1][num_midi_features*num_syllables_per_sentence:None])
else:
batch_meta_wrong.append(test[i-1][num_midi_features*num_syllables_per_sentence:None])
except:
break
try:
batch_songs_list = np.split(np.asarray(batch_songs), indices_or_sections=num_syllables_per_sentence, axis=1)
batch_songs_list = np.transpose(batch_songs_list, axes=(1, 0, 2))
batch_meta_list = np.split(np.asarray(batch_meta),
indices_or_sections=num_syllables_per_sentence*num_syllables_features, axis=1)
batch_meta_wrong_list = np.split(np.asarray(batch_meta_wrong),
indices_or_sections=num_syllables_per_sentence*num_syllables_features, axis=1)
except:
batch_songs_list = None
batch_meta = None
batch_meta_wrong = None
pointer = pointer + batch_size
return batch_songs_list, batch_meta, batch_meta_wrong, pointer
def get_batch_no_cond(batch_size, pointer, train, validate, test, num_midi_features, num_syllables_per_sentence,
num_syllables_features, part='train'):
batch_songs = []
if part == 'train':
batch_size = min(batch_size, np.shape(train[pointer:None])[0])
try:
for i in range(pointer, pointer+batch_size):
batch_songs.append(train[i][0:num_midi_features*num_syllables_per_sentence])
except:
batch_songs = None
elif part == 'validation':
batch_size = min(batch_size, np.shape(validate[pointer:None])[0])
try:
for i in range(pointer, pointer+batch_size):
batch_songs.append(validate[i][0:num_midi_features*num_syllables_per_sentence])
except:
batch_songs = None
else:
batch_size = min(batch_size, np.shape(test[pointer:None])[0])
try:
for i in range(pointer, pointer+batch_size):
batch_songs.append(test[i][0:num_midi_features*num_syllables_per_sentence])
except:
batch_songs = None
try:
batch_songs_list = np.split(np.asarray(batch_songs), indices_or_sections=num_syllables_per_sentence, axis=1)
batch_songs_list = np.transpose(batch_songs_list, axes=(1, 0, 2))
except:
batch_songs_list = None
pointer = pointer + batch_size
return batch_songs_list, pointer
def generate_fake_data(num_midi_features, num_syllables_features, num_syllables_per_sentence, num_songs,
dataset_type='Sentence level'):
labels = []
if dataset_type == 'Sentence level':
for i in range(0, num_syllables_per_sentence):
labels.append('time' + str(i))
labels.append('duration' + str(i))
labels.append('freq' + str(i))
labels.append('velocity' + str(i))
for i in range(0, num_syllables_per_sentence):
for j in range(0, num_syllables_features):
labels.append('feature' + str(j) + 'syllable' + str(i))
print(num_songs,num_syllables_per_sentence,num_midi_features,num_syllables_features)
df = pd.DataFrame(120*np.random.random(size=(num_songs, num_syllables_per_sentence *
(num_midi_features + num_syllables_features))), columns=labels)
return df
def normalise_data(train, vali, test, low=-1, high=1):
""" Apply some sort of whitening procedure
"""
mean = np.mean(np.vstack([train, vali]), axis=(0, 1))
std = np.std(np.vstack([train-mean, vali-mean]), axis=(0, 1))
normalised_train = (train - mean)/std
normalised_vali = (vali - mean)/std
normalised_test = (test - mean)/std
return normalised_train, normalised_vali, normalised_test
def denormalise_data(data, mean, std):
denormalised_data = data*std + mean
return denormalised_data
def create_midi_pattern(sample, tones_per_cell, num_features_per_tone):
# Create and save midi pattern
output_ticks_per_quarter_note = 384.0
midi_pattern = midi.Pattern([], resolution=int(output_ticks_per_quarter_note))
cur_track = midi.Track([])
cur_track.append(midi.events.SetTempoEvent(tick=0, bpm=100))
future_events = {}
last_event_tick = 0
ticks_to_this_tone = 0.0
song_events_absolute_ticks = []
abs_tick_note_beginning = 0.0
for frame in sample:
abs_tick_note_beginning += frame[0]
for subframe in range(tones_per_cell):
offset = subframe * num_features_per_tone
tick_len = int(round(frame[offset + 1]))
freq = frame[offset + 2]
velocity = min(int(round(frame[offset + 3])), 127)
d = freq_to_tone(freq)
if d is not None and velocity > 0 and tick_len > 0:
# range-check with preserved tone, changed one octave:
tone = d['tone']
while tone < 0:
tone += 12
while tone > 127:
tone -= 12
pitch_wheel = cents_to_pitchwheel_units(d['cents'])
song_events_absolute_ticks.append((abs_tick_note_beginning,
midi.events.NoteOnEvent(
tick=0,
velocity=velocity,
pitch=tone)))
song_events_absolute_ticks.append((abs_tick_note_beginning + tick_len,
midi.events.NoteOffEvent(
tick=0,
velocity=0,
pitch=tone)))
song_events_absolute_ticks.sort(key=lambda e: e[0])
abs_tick_note_beginning = 0.0
for abs_tick, event in song_events_absolute_ticks:
rel_tick = abs_tick - abs_tick_note_beginning
event.tick = int(round(rel_tick))
cur_track.append(event)
abs_tick_note_beginning = abs_tick
cur_track.append(midi.EndOfTrackEvent(tick=int(output_ticks_per_quarter_note)))
midi_pattern.append(cur_track)
return midi_pattern
def discretize(sample):
dist = np.inf
authorized_values_pitch = range(127)
authorized_values_duration = [0.25, 0.5, 0.75, 1., 1.5, 2., 3., 4., 6., 8., 16., 32.]
authorized_values_rest = [0., 1., 2., 4., 8., 16., 32.]
discretized_sample = np.zeros(shape=np.shape(sample))
discretized_sample_arrays = []
for i in range(len(sample)):
for j in range(0, len(authorized_values_pitch)):
if (sample[i][0] - authorized_values_pitch[j]) ** 2 < dist:
dist = (sample[i][0] - authorized_values_pitch[j]) ** 2
discretized_sample[i][0] = authorized_values_pitch[j]
dist = np.inf
for j in range(0, len(authorized_values_duration)):
if (sample[i][1] - authorized_values_duration[j]) ** 2 < dist:
dist = (sample[i][1] - authorized_values_duration[j]) ** 2
discretized_sample[i][1] = authorized_values_duration[j]
dist = np.inf
for j in range(0, len(authorized_values_rest)):
if (sample[i][2] - authorized_values_rest[j]) ** 2 < dist:
dist = (sample[i][2] - authorized_values_rest[j]) ** 2
discretized_sample[i][2] = authorized_values_rest[j]
dist = np.inf
discretized_sample_arrays.append(np.asarray(discretized_sample[i][:]))
return discretized_sample_arrays
def create_midi_pattern_from_discretized_data(discretized_sample):
new_midi = pretty_midi.PrettyMIDI()
voice = pretty_midi.Instrument(1) # It's here to change the used instruments !
tempo = 120
ActualTime = 0 # Time since the beginning of the song, in seconds
for i in range(0,len(discretized_sample)):
length = discretized_sample[i][1] * 60 / tempo # Conversion Duration to Time
if i < len(discretized_sample) - 1:
gap = discretized_sample[i + 1][2] * 60 / tempo
else:
gap = 0 # The Last element doesn't have a gap
note = pretty_midi.Note(velocity=100, pitch=int(discretized_sample[i][0]), start=ActualTime,
end=ActualTime + length)
voice.notes.append(note)
ActualTime += length + gap # Update of the time
new_midi.instruments.append(voice)
return new_midi
def mmd2(x, y, sigma=1):
print(len(x))
print(len(y))
var_x = var_u(x, sigma)
var_y = var_u(y, sigma)
covar = covar_u(x, y, sigma)
print("MMD2_x", var_x)
print("MMD2_y", var_y)
print("MMD2_xy", covar)
return var_x - covar + var_y
def var_u(x, sigma):
var_u = 0
n = len(x)
for i in range(0, n):
for j in range(0, n):
if i != j:
var_u += rbf(x[i], x[j], sigma)
n = np.double(len(x))
print(1./(n*(n-1)) * var_u)
return 1./(n*(n-1)) * var_u
def covar_u(x, y, sigma):
n = len(x)
m = len(y)
covar_u = 0
for i in range(0, n):
for j in range(0, m):
covar_u += rbf(x[i], y[j], sigma)
n = np.double(n)
m = np.double(m)
return 2./(m * n) * covar_u
def rbf(x, y, sigma):
frob = (sum((x - y) ** 2))
return np.exp(-frob/((2*sigma)**2))
def print_model_stats(model_stats, num_songs):
model_stats['stats_scale_tot'] = model_stats['stats_scale_tot'] / num_songs
model_stats['stats_repetitions_2_tot'] = model_stats['stats_repetitions_2_tot'] / num_songs
model_stats['stats_repetitions_3_tot'] = model_stats['stats_repetitions_3_tot'] / num_songs
model_stats['stats_span_tot'] = model_stats['stats_span_tot'] / num_songs
model_stats['stats_unique_tones_tot'] = model_stats['stats_unique_tones_tot'] / num_songs
model_stats['stats_avg_rest_tot'] = model_stats['stats_avg_rest_tot'] / num_songs
model_stats['num_of_null_rest_tot'] = model_stats['num_of_null_rest_tot'] / num_songs
model_stats['songlength_tot'] = model_stats['songlength_tot'] / num_songs
print('Average scale score :', model_stats['stats_scale_tot'])
print('Average repetitions of len 2 :', model_stats['stats_repetitions_2_tot'])
print('Average repetitions of len 3 :', model_stats['stats_repetitions_3_tot'])
print('Average span (in tones) :', model_stats['stats_span_tot'])
print('Average unique tones :', model_stats['stats_unique_tones_tot'])
print('Average rest over all notes :', model_stats['stats_avg_rest_tot'])
print('Average number of null rests :', model_stats['num_of_null_rest_tot'])
print('Average songlength :', model_stats['songlength_tot'])
print("*************************************************************************************************\n")
print("BEST STATS OVER {} GENERATED SONGS============================================================\n"
.format(num_songs))
print('Best scale score :', model_stats['best_scale_score'])
print('Highest num of repetition(s) of len 2 :', model_stats['best_repetitions_2'])
print('Highest num of repetition(s) of len 3 :', model_stats['best_repetitions_3'])
print('Number of perfect scale scores :', model_stats['num_perfect_scale'])
print('Number of "good" songs :', model_stats['num_good_songs'])
print("*************************************************************************************************\n")
def get_model_stats(model_stats, num_songs):
average_model_stats = {}
average_model_stats['stats_scale'] = model_stats['stats_scale_tot'] / num_songs
average_model_stats['stats_repetitions_2'] = model_stats['stats_repetitions_2_tot'] / num_songs
average_model_stats['stats_repetitions_3'] = model_stats['stats_repetitions_3_tot'] / num_songs
average_model_stats['stats_span'] = model_stats['stats_span_tot'] / num_songs
average_model_stats['stats_unique_tones'] = model_stats['stats_unique_tones_tot'] / num_songs
average_model_stats['stats_avg_rest'] = model_stats['stats_avg_rest_tot'] / num_songs
average_model_stats['num_of_null_rest'] = model_stats['num_of_null_rest_tot'] / num_songs
average_model_stats['songlength'] = model_stats['songlength_tot'] / num_songs
return average_model_stats
def discretize_length(sample):
dist = np.inf
authorized_values_duration = [0.25, 0.5, 0.75, 1., 1.5, 2., 3., 4., 6., 8., 16., 32.]
discretized_sample = np.zeros(shape=np.shape(sample))
discretized_sample_arrays = []
for i in range(len(sample)):
for j in range(0, len(authorized_values_duration)):
if (sample[i][0] - authorized_values_duration[j]) ** 2 < dist:
dist = (sample[i][0] - authorized_values_duration[j]) ** 2
discretized_sample[i][0] = authorized_values_duration[j]
dist = np.inf
discretized_sample_arrays.append(np.asarray(discretized_sample[i][:]))
return discretized_sample_arrays
def discretize_rest(sample):
dist = np.inf
authorized_values_rest = [0, 1, 2, 4, 8, 16, 32]
discretized_sample = np.zeros(shape=np.shape(sample))
discretized_sample_arrays = []
for i in range(len(sample)):
for j in range(0, len(authorized_values_rest)):
if (sample[i][0] - authorized_values_rest[j]) ** 2 < dist:
dist = (sample[i][0] - authorized_values_rest[j]) ** 2
discretized_sample[i][0] = authorized_values_rest[j]
dist = np.inf
discretized_sample_arrays.append(np.asarray(discretized_sample[i][:]))
return discretized_sample_arrays
def discretize_pitch(sample):
dist = np.inf
authorized_values_pitch = range(127)
discretized_sample = np.zeros(shape=np.shape(sample))
discretized_sample_arrays = []
for i in range(len(sample)):
for j in range(0, len(authorized_values_pitch)):
if (sample[i][0] - authorized_values_pitch[j]) ** 2 < dist:
dist = (sample[i][0] - authorized_values_pitch[j]) ** 2
discretized_sample[i][0] = authorized_values_pitch[j]
dist = np.inf
discretized_sample_arrays.append(np.asarray(discretized_sample[i][:]))
return discretized_sample_arrays