projectlosangeles commited on
Commit
142b9a5
1 Parent(s): cec716f

Upload 6 files

Browse files
Master_MIDI_Dataset_GPU_Search_and_Filter.ipynb ADDED
@@ -0,0 +1,574 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "markdown",
5
+ "metadata": {
6
+ "gradient": {
7
+ "editing": false,
8
+ "id": "ac5a4cf0-d9d2-47b5-9633-b53f8d99a4d2",
9
+ "kernelId": ""
10
+ },
11
+ "id": "SiTIpPjArIyr"
12
+ },
13
+ "source": [
14
+ "# Master MIDI Dataset GPU Search and Filter (ver. 2.0)\n",
15
+ "\n",
16
+ "***\n",
17
+ "\n",
18
+ "Powered by tegridy-tools: https://github.com/asigalov61/tegridy-tools\n",
19
+ "\n",
20
+ "***\n",
21
+ "\n",
22
+ "#### Project Los Angeles\n",
23
+ "\n",
24
+ "#### Tegridy Code 2024\n",
25
+ "\n",
26
+ "***"
27
+ ]
28
+ },
29
+ {
30
+ "cell_type": "markdown",
31
+ "metadata": {
32
+ "gradient": {
33
+ "editing": false,
34
+ "id": "fa0a611c-1803-42ae-bdf6-a49b5a4e781b",
35
+ "kernelId": ""
36
+ },
37
+ "id": "gOd93yV0sGd2"
38
+ },
39
+ "source": [
40
+ "# (SETUP ENVIRONMENT)"
41
+ ]
42
+ },
43
+ {
44
+ "cell_type": "markdown",
45
+ "source": [
46
+ "# ( GPU CHECK)"
47
+ ],
48
+ "metadata": {
49
+ "id": "0rMwKVc9FFRw"
50
+ }
51
+ },
52
+ {
53
+ "cell_type": "code",
54
+ "source": [
55
+ "# @title NVIDIA GPU Check\n",
56
+ "!nvidia-smi"
57
+ ],
58
+ "metadata": {
59
+ "cellView": "form",
60
+ "id": "dVSaUaEZFIip"
61
+ },
62
+ "execution_count": null,
63
+ "outputs": []
64
+ },
65
+ {
66
+ "cell_type": "markdown",
67
+ "source": [
68
+ "# (SETUP ENVIRONMENT)"
69
+ ],
70
+ "metadata": {
71
+ "id": "YRTt3Hx0FQeu"
72
+ }
73
+ },
74
+ {
75
+ "cell_type": "code",
76
+ "execution_count": null,
77
+ "metadata": {
78
+ "cellView": "form",
79
+ "gradient": {
80
+ "editing": false,
81
+ "id": "a1a45a91-d909-4fd4-b67a-5e16b971d179",
82
+ "kernelId": ""
83
+ },
84
+ "id": "fX12Yquyuihc"
85
+ },
86
+ "outputs": [],
87
+ "source": [
88
+ "#@title Install all dependencies (run only once per session)\n",
89
+ "\n",
90
+ "!git clone --depth 1 https://github.com/asigalov61/Los-Angeles-MIDI-Dataset\n",
91
+ "!pip install huggingface_hub"
92
+ ]
93
+ },
94
+ {
95
+ "cell_type": "code",
96
+ "execution_count": null,
97
+ "metadata": {
98
+ "gradient": {
99
+ "editing": false,
100
+ "id": "b8207b76-9514-4c07-95db-95a4742e52c5",
101
+ "kernelId": ""
102
+ },
103
+ "id": "z7n9vnKmug1J",
104
+ "cellView": "form"
105
+ },
106
+ "outputs": [],
107
+ "source": [
108
+ "#@title Import all needed modules\n",
109
+ "\n",
110
+ "print('Loading core modules... Please wait...')\n",
111
+ "\n",
112
+ "import os\n",
113
+ "import copy\n",
114
+ "from collections import Counter\n",
115
+ "import random\n",
116
+ "import pickle\n",
117
+ "from tqdm import tqdm\n",
118
+ "import pprint\n",
119
+ "import statistics\n",
120
+ "import shutil\n",
121
+ "\n",
122
+ "import cupy as cp\n",
123
+ "\n",
124
+ "from huggingface_hub import hf_hub_download\n",
125
+ "\n",
126
+ "print('Loading TMIDIX module...')\n",
127
+ "os.chdir('/content/Los-Angeles-MIDI-Dataset')\n",
128
+ "\n",
129
+ "import TMIDIX\n",
130
+ "\n",
131
+ "os.chdir('/content/')\n",
132
+ "\n",
133
+ "print('Creating IO dirs... Please wait...')\n",
134
+ "\n",
135
+ "if not os.path.exists('/content/Master-MIDI-Dataset'):\n",
136
+ " os.makedirs('/content/Master-MIDI-Dataset')\n",
137
+ "\n",
138
+ "if not os.path.exists('/content/Master-MIDI-Dataset'):\n",
139
+ " os.makedirs('/content/Master-MIDI-Dataset')\n",
140
+ "\n",
141
+ "if not os.path.exists('/content/Output-MIDI-Dataset'):\n",
142
+ " os.makedirs('/content/Output-MIDI-Dataset')\n",
143
+ "\n",
144
+ "print('Done!')\n",
145
+ "print('Enjoy! :)')"
146
+ ]
147
+ },
148
+ {
149
+ "cell_type": "markdown",
150
+ "metadata": {
151
+ "gradient": {
152
+ "editing": false,
153
+ "id": "20b8698a-0b4e-4fdb-ae49-24d063782e77",
154
+ "kernelId": ""
155
+ },
156
+ "id": "ObPxlEutsQBj"
157
+ },
158
+ "source": [
159
+ "# (PREP MAIN MIDI DATASET)"
160
+ ]
161
+ },
162
+ {
163
+ "cell_type": "code",
164
+ "source": [
165
+ "#@title Download Los Angeles MIDI Dataset\n",
166
+ "print('=' * 70)\n",
167
+ "print('Downloading Los Angeles MIDI Dataset...Please wait...')\n",
168
+ "print('=' * 70)\n",
169
+ "\n",
170
+ "hf_hub_download(repo_id='projectlosangeles/Los-Angeles-MIDI-Dataset',\n",
171
+ " filename='Los-Angeles-MIDI-Dataset-Ver-4-0-CC-BY-NC-SA.zip',\n",
172
+ " repo_type=\"dataset\",\n",
173
+ " local_dir='/content/Main-MIDI-Dataset',\n",
174
+ " local_dir_use_symlinks=False)\n",
175
+ "print('=' * 70)\n",
176
+ "print('Done! Enjoy! :)')\n",
177
+ "print('=' * 70)"
178
+ ],
179
+ "metadata": {
180
+ "cellView": "form",
181
+ "id": "7aItlhq9cRxZ"
182
+ },
183
+ "execution_count": null,
184
+ "outputs": []
185
+ },
186
+ {
187
+ "cell_type": "code",
188
+ "source": [
189
+ "#@title Unzip Los Angeles MIDI Dataset\n",
190
+ "%cd /content/Main-MIDI-Dataset/\n",
191
+ "\n",
192
+ "print('=' * 70)\n",
193
+ "print('Unzipping Los Angeles MIDI Dataset...Please wait...')\n",
194
+ "!unzip 'Los-Angeles-MIDI-Dataset-Ver-4-0-CC-BY-NC-SA.zip'\n",
195
+ "print('=' * 70)\n",
196
+ "\n",
197
+ "print('Done! Enjoy! :)')\n",
198
+ "print('=' * 70)\n",
199
+ "%cd /content/"
200
+ ],
201
+ "metadata": {
202
+ "cellView": "form",
203
+ "id": "zMF4vdMNDYYg"
204
+ },
205
+ "execution_count": null,
206
+ "outputs": []
207
+ },
208
+ {
209
+ "cell_type": "code",
210
+ "source": [
211
+ "#@title Create Los Angeles MIDI Dataset files list\n",
212
+ "print('=' * 70)\n",
213
+ "print('Creating dataset files list...')\n",
214
+ "dataset_addr = \"/content/Main-MIDI-Dataset/MIDIs\"\n",
215
+ "\n",
216
+ "# os.chdir(dataset_addr)\n",
217
+ "filez = list()\n",
218
+ "for (dirpath, dirnames, filenames) in os.walk(dataset_addr):\n",
219
+ " filez += [os.path.join(dirpath, file) for file in filenames]\n",
220
+ "\n",
221
+ "if filez == []:\n",
222
+ " print('Could not find any MIDI files. Please check Dataset dir...')\n",
223
+ " print('=' * 70)\n",
224
+ "\n",
225
+ "print('=' * 70)\n",
226
+ "print('Randomizing file list...')\n",
227
+ "random.shuffle(filez)\n",
228
+ "print('=' * 70)\n",
229
+ "\n",
230
+ "LAMD_files_list = []\n",
231
+ "\n",
232
+ "for f in tqdm(filez):\n",
233
+ " LAMD_files_list.append([f.split('/')[-1].split('.mid')[0], f])\n",
234
+ "print('Done!')\n",
235
+ "print('=' * 70)"
236
+ ],
237
+ "metadata": {
238
+ "cellView": "form",
239
+ "id": "btrUDk8MDfdw"
240
+ },
241
+ "execution_count": null,
242
+ "outputs": []
243
+ },
244
+ {
245
+ "cell_type": "code",
246
+ "source": [
247
+ "#@title Load Los Angeles MIDI Dataset Signatures Data\n",
248
+ "\n",
249
+ "print('=' * 70)\n",
250
+ "print('Loading LAMDa Signatures Data...')\n",
251
+ "sigs_data = pickle.load(open('/content/Main-MIDI-Dataset/SIGNATURES_DATA/LAMDa_SIGNATURES_DATA.pickle', 'rb'))\n",
252
+ "print('=' * 70)\n",
253
+ "\n",
254
+ "print('Prepping signatures...')\n",
255
+ "print('=' * 70)\n",
256
+ "\n",
257
+ "random.shuffle(sigs_data)\n",
258
+ "\n",
259
+ "signatures_file_names = []\n",
260
+ "sigs_matrixes = [ [0]*(len(TMIDIX.ALL_CHORDS)+128) for i in range(len(sigs_data))]\n",
261
+ "\n",
262
+ "idx = 0\n",
263
+ "for s in tqdm(sigs_data):\n",
264
+ "\n",
265
+ " signatures_file_names.append(s[0])\n",
266
+ "\n",
267
+ " counts_sum = sum([c[1] for c in s[1]])\n",
268
+ "\n",
269
+ " for ss in s[1]:\n",
270
+ " sigs_matrixes[idx][ss[0]] = ss[1] / counts_sum\n",
271
+ "\n",
272
+ " idx += 1\n",
273
+ "\n",
274
+ "print('=' * 70)\n",
275
+ "print('Loading signatures...')\n",
276
+ "print('=' * 70)\n",
277
+ "\n",
278
+ "signatures_data = cp.array(sigs_matrixes)\n",
279
+ "\n",
280
+ "print('Done!')\n",
281
+ "print('=' * 70)"
282
+ ],
283
+ "metadata": {
284
+ "id": "Mv-pjxbrIqi2",
285
+ "cellView": "form"
286
+ },
287
+ "execution_count": null,
288
+ "outputs": []
289
+ },
290
+ {
291
+ "cell_type": "markdown",
292
+ "source": [
293
+ "# (SEARCH AND FILTER)\n",
294
+ "\n",
295
+ "### DO NOT FORGET TO UPLOAD YOUR MASTER DATASET TO \"Master-MIDI-Dataset\" FOLDER"
296
+ ],
297
+ "metadata": {
298
+ "id": "iaeqXuIHI0_T"
299
+ }
300
+ },
301
+ {
302
+ "cell_type": "code",
303
+ "source": [
304
+ "#@title Master MIDI Dataset Search and Filter\n",
305
+ "\n",
306
+ "#@markdown NOTE: You can stop the search at any time to render partial results\n",
307
+ "\n",
308
+ "number_of_top_matches_MIDIs_to_collect = 20 #@param {type:\"slider\", min:5, max:50, step:1}\n",
309
+ "search_matching_type = \"ratios\" # @param [\"ratios\", \"distances\"]\n",
310
+ "distances_norm_order = 3 # @param {type:\"slider\", min:1, max:10, step:1}\n",
311
+ "maximum_match_ratio_to_search_for = 0.999 #@param {type:\"slider\", min:0, max:1, step:0.001}\n",
312
+ "\n",
313
+ "print('=' * 70)\n",
314
+ "print('Master MIDI Dataset GPU Search and Filter')\n",
315
+ "print('=' * 70)\n",
316
+ "\n",
317
+ "###########\n",
318
+ "\n",
319
+ "print('Loading MIDI files...')\n",
320
+ "print('This may take a while on a large dataset in particular.')\n",
321
+ "\n",
322
+ "dataset_addr = \"/content/Master-MIDI-Dataset\"\n",
323
+ "\n",
324
+ "filez = list()\n",
325
+ "\n",
326
+ "for (dirpath, dirnames, filenames) in os.walk(dataset_addr):\n",
327
+ " for file in filenames:\n",
328
+ " if file.endswith(('.mid', '.midi', '.kar')):\n",
329
+ " filez.append(os.path.join(dirpath, file))\n",
330
+ "\n",
331
+ "print('=' * 70)\n",
332
+ "\n",
333
+ "if filez:\n",
334
+ "\n",
335
+ " print('Randomizing file list...')\n",
336
+ " random.shuffle(filez)\n",
337
+ " print('=' * 70)\n",
338
+ "\n",
339
+ " ###################\n",
340
+ "\n",
341
+ " if not os.path.exists('/content/Output-MIDI-Dataset'):\n",
342
+ " os.makedirs('/content/Output-MIDI-Dataset')\n",
343
+ "\n",
344
+ " ###################\n",
345
+ "\n",
346
+ " input_files_count = 0\n",
347
+ " files_count = 0\n",
348
+ "\n",
349
+ " for f in filez:\n",
350
+ " try:\n",
351
+ "\n",
352
+ " input_files_count += 1\n",
353
+ "\n",
354
+ " fn = os.path.basename(f)\n",
355
+ " fn1 = os.path.splitext(fn)[0]\n",
356
+ " ext = os.path.splitext(f)[1]\n",
357
+ "\n",
358
+ " print('Processing MIDI File #', files_count+1, 'out of', len(filez))\n",
359
+ " print('MIDI file name', fn)\n",
360
+ " print('-' * 70)\n",
361
+ "\n",
362
+ " #=======================================================\n",
363
+ "\n",
364
+ " raw_score = TMIDIX.midi2single_track_ms_score(open(f, 'rb').read())\n",
365
+ " escore = TMIDIX.advanced_score_processor(raw_score, return_score_analysis=False, return_enhanced_score_notes=True)[0]\n",
366
+ "\n",
367
+ " for e in escore:\n",
368
+ " e[1] = int(e[1] / 16)\n",
369
+ " e[2] = int(e[2] / 16)\n",
370
+ "\n",
371
+ " src_sigs = []\n",
372
+ "\n",
373
+ " for i in range(-6, 6):\n",
374
+ "\n",
375
+ " escore_copy = copy.deepcopy(escore)\n",
376
+ "\n",
377
+ " for e in escore_copy:\n",
378
+ " e[4] += i\n",
379
+ "\n",
380
+ " cscore = TMIDIX.chordify_score([1000, escore_copy])\n",
381
+ "\n",
382
+ " sig = []\n",
383
+ "\n",
384
+ " for c in cscore:\n",
385
+ "\n",
386
+ " pitches = sorted(set([p[4] for p in c if p[3] != 9]))\n",
387
+ "\n",
388
+ " if pitches:\n",
389
+ " if len(pitches) > 1:\n",
390
+ " tones_chord = sorted(set([p % 12 for p in pitches]))\n",
391
+ " checked_tones_chord = TMIDIX.check_and_fix_tones_chord(tones_chord)\n",
392
+ "\n",
393
+ " sig_token = TMIDIX.ALL_CHORDS.index(checked_tones_chord) + 128\n",
394
+ "\n",
395
+ " elif len(pitches) == 1:\n",
396
+ " sig_token = pitches[0]\n",
397
+ "\n",
398
+ " sig.append(sig_token)\n",
399
+ "\n",
400
+ " fsig = [list(v) for v in Counter(sig).most_common()]\n",
401
+ "\n",
402
+ " src_sig_mat = [0] * (len(TMIDIX.ALL_CHORDS)+128)\n",
403
+ "\n",
404
+ " counts_sum = sum([c[1] for c in fsig])\n",
405
+ "\n",
406
+ " for s in fsig:\n",
407
+ "\n",
408
+ " src_sig_mat[s[0]] = s[1] / counts_sum\n",
409
+ "\n",
410
+ " src_sigs.append(src_sig_mat)\n",
411
+ "\n",
412
+ " src_signatures = cp.stack(cp.array(src_sigs))\n",
413
+ "\n",
414
+ " #=======================================================\n",
415
+ "\n",
416
+ " print('Searching for matches...Please wait...')\n",
417
+ " print('-' * 70)\n",
418
+ "\n",
419
+ " lower_threshold = 0.0\n",
420
+ " upper_threshold = maximum_match_ratio_to_search_for\n",
421
+ " filter_size = number_of_top_matches_MIDIs_to_collect\n",
422
+ "\n",
423
+ " final_ratios = []\n",
424
+ "\n",
425
+ " avg_idxs = []\n",
426
+ "\n",
427
+ " all_filtered_means = []\n",
428
+ " all_filtered_idxs = []\n",
429
+ " all_filtered_tvs = []\n",
430
+ "\n",
431
+ " tv_idx = -6\n",
432
+ "\n",
433
+ " for target_sig in tqdm(src_signatures):\n",
434
+ "\n",
435
+ " if search_matching_type == 'ratios':\n",
436
+ "\n",
437
+ " ratios = cp.where(target_sig != 0, cp.divide(cp.minimum(signatures_data, target_sig), cp.maximum(signatures_data, target_sig)), 0)\n",
438
+ " max_comp_lengths = cp.maximum(cp.repeat(cp.sum(target_sig != 0), signatures_data.shape[0]), cp.sum(signatures_data != 0, axis=1))\n",
439
+ "\n",
440
+ " results = cp.divide(cp.sum(ratios, axis=1), max_comp_lengths)\n",
441
+ "\n",
442
+ " elif search_matching_type == 'distances':\n",
443
+ "\n",
444
+ " distances = cp.power(cp.sum(cp.power(cp.abs(signatures_data - target_sig), distances_norm_order), axis=1), 1 / distances_norm_order)\n",
445
+ "\n",
446
+ " results = cp.max(distances) - distances\n",
447
+ "\n",
448
+ " unique_means = cp.unique(results)\n",
449
+ " sorted_means = cp.sort(unique_means)[::-1]\n",
450
+ "\n",
451
+ " filtered_means = sorted_means[(sorted_means >= lower_threshold) & (sorted_means <= upper_threshold)][:filter_size]\n",
452
+ "\n",
453
+ " filtered_idxs = cp.where(cp.in1d(results, filtered_means))[0]\n",
454
+ "\n",
455
+ " all_filtered_means.extend(results[cp.in1d(results, filtered_means)].tolist())\n",
456
+ "\n",
457
+ " all_filtered_idxs.extend(filtered_idxs.tolist())\n",
458
+ "\n",
459
+ " filtered_tvs = [tv_idx] * filtered_idxs.shape[0]\n",
460
+ "\n",
461
+ " all_filtered_tvs.extend(filtered_tvs)\n",
462
+ "\n",
463
+ " tv_idx += 1\n",
464
+ "\n",
465
+ " filtered_results = sorted(zip(all_filtered_means, all_filtered_idxs, all_filtered_tvs), key=lambda x: x[0], reverse=True)[:filter_size]\n",
466
+ "\n",
467
+ " #=======================================================\n",
468
+ "\n",
469
+ " print('Done!')\n",
470
+ " print('-' * 70)\n",
471
+ " print('Max match ratio:', filtered_results[0][0])\n",
472
+ " print('Max match transpose value:', filtered_results[0][2])\n",
473
+ " print('Max match signature index:', filtered_results[0][1])\n",
474
+ " print('Max match file name:', signatures_file_names[filtered_results[0][1]])\n",
475
+ " print('-' * 70)\n",
476
+ " print('Copying max ratios MIDIs...')\n",
477
+ "\n",
478
+ " for fr in filtered_results:\n",
479
+ "\n",
480
+ " max_ratio_index = fr[1]\n",
481
+ "\n",
482
+ " ffn = signatures_file_names[fr[1]]\n",
483
+ " ffn_idx = [y[0] for y in LAMD_files_list].index(ffn)\n",
484
+ "\n",
485
+ " ff = LAMD_files_list[ffn_idx][1]\n",
486
+ "\n",
487
+ " #=======================================================\n",
488
+ "\n",
489
+ " dir_str = str(fn1)\n",
490
+ " copy_path = '/content/Output-MIDI-Dataset/'+dir_str\n",
491
+ " if not os.path.exists(copy_path):\n",
492
+ " os.mkdir(copy_path)\n",
493
+ "\n",
494
+ " fff = str(fr[0] * 100) + '_' + str(fr[2]) + '_' + ffn + '.mid'\n",
495
+ "\n",
496
+ " shutil.copy2(ff, copy_path+'/'+fff)\n",
497
+ "\n",
498
+ " shutil.copy2(f, copy_path+'/'+fn)\n",
499
+ "\n",
500
+ " #======================================================='''\n",
501
+ " print('Done!')\n",
502
+ " print('=' * 70)\n",
503
+ "\n",
504
+ " #=======================================================\n",
505
+ "\n",
506
+ " # Processed files counter\n",
507
+ " files_count += 1\n",
508
+ "\n",
509
+ " except KeyboardInterrupt:\n",
510
+ " print('Quitting...')\n",
511
+ " print('Total number of processed MIDI files', files_count)\n",
512
+ " print('=' * 70)\n",
513
+ " break\n",
514
+ "\n",
515
+ " except Exception as ex:\n",
516
+ " print('WARNING !!!')\n",
517
+ " print('=' * 70)\n",
518
+ " print('Bad file:', f)\n",
519
+ " print('Error detected:', ex)\n",
520
+ " print('=' * 70)\n",
521
+ " continue\n",
522
+ "\n",
523
+ " print('Total number of processed MIDI files', files_count)\n",
524
+ " print('=' * 70)\n",
525
+ "\n",
526
+ "else:\n",
527
+ " print('Could not find any MIDI files. Please check Dataset dir...')\n",
528
+ " print('=' * 70)"
529
+ ],
530
+ "metadata": {
531
+ "cellView": "form",
532
+ "id": "M0JWCPzBGNvh"
533
+ },
534
+ "execution_count": null,
535
+ "outputs": []
536
+ },
537
+ {
538
+ "cell_type": "markdown",
539
+ "metadata": {
540
+ "id": "YzCMd94Tu_gz"
541
+ },
542
+ "source": [
543
+ "# Congrats! You did it! :)"
544
+ ]
545
+ }
546
+ ],
547
+ "metadata": {
548
+ "colab": {
549
+ "private_outputs": true,
550
+ "provenance": [],
551
+ "gpuType": "T4",
552
+ "machine_shape": "hm"
553
+ },
554
+ "kernelspec": {
555
+ "display_name": "Python 3",
556
+ "name": "python3"
557
+ },
558
+ "language_info": {
559
+ "codemirror_mode": {
560
+ "name": "ipython",
561
+ "version": 3
562
+ },
563
+ "file_extension": ".py",
564
+ "mimetype": "text/x-python",
565
+ "name": "python",
566
+ "nbconvert_exporter": "python",
567
+ "pygments_lexer": "ipython3",
568
+ "version": "3.9.7"
569
+ },
570
+ "accelerator": "GPU"
571
+ },
572
+ "nbformat": 4,
573
+ "nbformat_minor": 0
574
+ }
TMIDIX.py CHANGED
@@ -3852,7 +3852,8 @@ ALL_CHORDS = [[0], [7], [5], [9], [2], [4], [11], [10], [8], [6], [3], [1], [0,
3852
  [2, 5, 7, 9, 11], [1, 3, 5, 7, 10], [0, 2, 4, 7, 10], [1, 3, 5, 7, 9],
3853
  [1, 3, 5, 9, 11], [1, 5, 7, 9, 11], [1, 3, 7, 9, 11], [3, 5, 7, 9, 11],
3854
  [2, 4, 6, 8, 10], [0, 4, 6, 8, 10], [0, 2, 6, 8, 10], [1, 3, 5, 7, 11],
3855
- [0, 2, 4, 8, 10], [0, 2, 4, 6, 8], [0, 2, 4, 6, 10]]
 
3856
 
3857
  def find_exact_match_variable_length(list_of_lists, target_list, uncertain_indices):
3858
  # Infer possible values for each uncertain index
@@ -3981,7 +3982,7 @@ def analyze_score_pitches(score, channels_to_analyze=[0]):
3981
 
3982
  ###################################################################################
3983
 
3984
- ALL_CHORDS_GROUPED = [
3985
  [[0, 2, 5, 7, 10], [0, 2, 4, 7, 9], [0, 2, 5, 7, 9], [1, 4, 6, 9, 11],
3986
  [1, 3, 6, 8, 11], [1, 3, 6, 8, 10], [1, 4, 6, 8, 11], [1, 3, 5, 8, 10],
3987
  [2, 4, 6, 9, 11], [2, 4, 7, 9, 11], [0, 3, 5, 7, 10], [0, 3, 5, 8, 10],
@@ -4510,6 +4511,33 @@ def ascii_text_words_counter(ascii_text):
4510
 
4511
  ###################################################################################
4512
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4513
  # This is the end of the TMIDI X Python module
4514
 
4515
  ###################################################################################
 
3852
  [2, 5, 7, 9, 11], [1, 3, 5, 7, 10], [0, 2, 4, 7, 10], [1, 3, 5, 7, 9],
3853
  [1, 3, 5, 9, 11], [1, 5, 7, 9, 11], [1, 3, 7, 9, 11], [3, 5, 7, 9, 11],
3854
  [2, 4, 6, 8, 10], [0, 4, 6, 8, 10], [0, 2, 6, 8, 10], [1, 3, 5, 7, 11],
3855
+ [0, 2, 4, 8, 10], [0, 2, 4, 6, 8], [0, 2, 4, 6, 10], [0, 2, 4, 6, 8, 10],
3856
+ [1, 3, 5, 7, 9, 11]]
3857
 
3858
  def find_exact_match_variable_length(list_of_lists, target_list, uncertain_indices):
3859
  # Infer possible values for each uncertain index
 
3982
 
3983
  ###################################################################################
3984
 
3985
+ ALL_CHORDS_GROUPED = [[[1, 3, 5, 7, 9, 11], [0, 2, 4, 6, 8, 10]],
3986
  [[0, 2, 5, 7, 10], [0, 2, 4, 7, 9], [0, 2, 5, 7, 9], [1, 4, 6, 9, 11],
3987
  [1, 3, 6, 8, 11], [1, 3, 6, 8, 10], [1, 4, 6, 8, 11], [1, 3, 5, 8, 10],
3988
  [2, 4, 6, 9, 11], [2, 4, 7, 9, 11], [0, 3, 5, 7, 10], [0, 3, 5, 8, 10],
 
4511
 
4512
  ###################################################################################
4513
 
4514
+ def check_and_fix_tones_chord(tones_chord):
4515
+
4516
+ lst = tones_chord
4517
+
4518
+ if len(lst) == 2:
4519
+ if lst[1] - lst[0] == 1:
4520
+ return [lst[-1]]
4521
+ else:
4522
+ if 0 in lst and 11 in lst:
4523
+ lst.remove(0)
4524
+ return lst
4525
+
4526
+ non_consecutive = [lst[0]]
4527
+
4528
+ if len(lst) > 2:
4529
+ for i in range(1, len(lst) - 1):
4530
+ if lst[i-1] + 1 != lst[i] and lst[i] + 1 != lst[i+1]:
4531
+ non_consecutive.append(lst[i])
4532
+ non_consecutive.append(lst[-1])
4533
+
4534
+ if 0 in non_consecutive and 11 in non_consecutive:
4535
+ non_consecutive.remove(0)
4536
+
4537
+ return non_consecutive
4538
+
4539
+ ###################################################################################
4540
+
4541
  # This is the end of the TMIDI X Python module
4542
 
4543
  ###################################################################################
master_midi_dataset_gpu_search_and_filter.py ADDED
@@ -0,0 +1,399 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # -*- coding: utf-8 -*-
2
+ """Master_MIDI_Dataset_GPU_Search_and_Filter.ipynb
3
+
4
+ Automatically generated by Colaboratory.
5
+
6
+ Original file is located at
7
+ https://colab.research.google.com/github/asigalov61/Los-Angeles-MIDI-Dataset/blob/main/Extras/Master_MIDI_Dataset_GPU_Search_and_Filter.ipynb
8
+
9
+ # Master MIDI Dataset GPU Search and Filter (ver. 2.0)
10
+
11
+ ***
12
+
13
+ Powered by tegridy-tools: https://github.com/asigalov61/tegridy-tools
14
+
15
+ ***
16
+
17
+ #### Project Los Angeles
18
+
19
+ #### Tegridy Code 2024
20
+
21
+ ***
22
+
23
+ # (SETUP ENVIRONMENT)
24
+
25
+ # ( GPU CHECK)
26
+ """
27
+
28
+ # @title NVIDIA GPU Check
29
+ !nvidia-smi
30
+
31
+ """# (SETUP ENVIRONMENT)"""
32
+
33
+ #@title Install all dependencies (run only once per session)
34
+
35
+ !git clone --depth 1 https://github.com/asigalov61/Los-Angeles-MIDI-Dataset
36
+ !pip install huggingface_hub
37
+
38
+ #@title Import all needed modules
39
+
40
+ print('Loading core modules... Please wait...')
41
+
42
+ import os
43
+ import copy
44
+ from collections import Counter
45
+ import random
46
+ import pickle
47
+ from tqdm import tqdm
48
+ import pprint
49
+ import statistics
50
+ import shutil
51
+
52
+ import cupy as cp
53
+
54
+ from huggingface_hub import hf_hub_download
55
+
56
+ print('Loading TMIDIX module...')
57
+ os.chdir('/content/Los-Angeles-MIDI-Dataset')
58
+
59
+ import TMIDIX
60
+
61
+ os.chdir('/content/')
62
+
63
+ print('Creating IO dirs... Please wait...')
64
+
65
+ if not os.path.exists('/content/Master-MIDI-Dataset'):
66
+ os.makedirs('/content/Master-MIDI-Dataset')
67
+
68
+ if not os.path.exists('/content/Master-MIDI-Dataset'):
69
+ os.makedirs('/content/Master-MIDI-Dataset')
70
+
71
+ if not os.path.exists('/content/Output-MIDI-Dataset'):
72
+ os.makedirs('/content/Output-MIDI-Dataset')
73
+
74
+ print('Done!')
75
+ print('Enjoy! :)')
76
+
77
+ """# (PREP MAIN MIDI DATASET)"""
78
+
79
+ #@title Download Los Angeles MIDI Dataset
80
+ print('=' * 70)
81
+ print('Downloading Los Angeles MIDI Dataset...Please wait...')
82
+ print('=' * 70)
83
+
84
+ hf_hub_download(repo_id='projectlosangeles/Los-Angeles-MIDI-Dataset',
85
+ filename='Los-Angeles-MIDI-Dataset-Ver-4-0-CC-BY-NC-SA.zip',
86
+ repo_type="dataset",
87
+ local_dir='/content/Main-MIDI-Dataset',
88
+ local_dir_use_symlinks=False)
89
+ print('=' * 70)
90
+ print('Done! Enjoy! :)')
91
+ print('=' * 70)
92
+
93
+ # Commented out IPython magic to ensure Python compatibility.
94
+ #@title Unzip Los Angeles MIDI Dataset
95
+ # %cd /content/Main-MIDI-Dataset/
96
+
97
+ print('=' * 70)
98
+ print('Unzipping Los Angeles MIDI Dataset...Please wait...')
99
+ !unzip 'Los-Angeles-MIDI-Dataset-Ver-4-0-CC-BY-NC-SA.zip'
100
+ print('=' * 70)
101
+
102
+ print('Done! Enjoy! :)')
103
+ print('=' * 70)
104
+ # %cd /content/
105
+
106
+ #@title Create Los Angeles MIDI Dataset files list
107
+ print('=' * 70)
108
+ print('Creating dataset files list...')
109
+ dataset_addr = "/content/Main-MIDI-Dataset/MIDIs"
110
+
111
+ # os.chdir(dataset_addr)
112
+ filez = list()
113
+ for (dirpath, dirnames, filenames) in os.walk(dataset_addr):
114
+ filez += [os.path.join(dirpath, file) for file in filenames]
115
+
116
+ if filez == []:
117
+ print('Could not find any MIDI files. Please check Dataset dir...')
118
+ print('=' * 70)
119
+
120
+ print('=' * 70)
121
+ print('Randomizing file list...')
122
+ random.shuffle(filez)
123
+ print('=' * 70)
124
+
125
+ LAMD_files_list = []
126
+
127
+ for f in tqdm(filez):
128
+ LAMD_files_list.append([f.split('/')[-1].split('.mid')[0], f])
129
+ print('Done!')
130
+ print('=' * 70)
131
+
132
+ #@title Load Los Angeles MIDI Dataset Signatures Data
133
+
134
+ print('=' * 70)
135
+ print('Loading LAMDa Signatures Data...')
136
+ sigs_data = pickle.load(open('/content/Main-MIDI-Dataset/SIGNATURES_DATA/LAMDa_SIGNATURES_DATA.pickle', 'rb'))
137
+ print('=' * 70)
138
+
139
+ print('Prepping signatures...')
140
+ print('=' * 70)
141
+
142
+ random.shuffle(sigs_data)
143
+
144
+ signatures_file_names = []
145
+ sigs_matrixes = [ [0]*(len(TMIDIX.ALL_CHORDS)+128) for i in range(len(sigs_data))]
146
+
147
+ idx = 0
148
+ for s in tqdm(sigs_data):
149
+
150
+ signatures_file_names.append(s[0])
151
+
152
+ counts_sum = sum([c[1] for c in s[1]])
153
+
154
+ for ss in s[1]:
155
+ sigs_matrixes[idx][ss[0]] = ss[1] / counts_sum
156
+
157
+ idx += 1
158
+
159
+ print('=' * 70)
160
+ print('Loading signatures...')
161
+ print('=' * 70)
162
+
163
+ signatures_data = cp.array(sigs_matrixes)
164
+
165
+ print('Done!')
166
+ print('=' * 70)
167
+
168
+ """# (SEARCH AND FILTER)
169
+
170
+ ### DO NOT FORGET TO UPLOAD YOUR MASTER DATASET TO "Master-MIDI-Dataset" FOLDER
171
+ """
172
+
173
+ #@title Master MIDI Dataset Search and Filter
174
+
175
+ #@markdown NOTE: You can stop the search at any time to render partial results
176
+
177
+ number_of_top_matches_MIDIs_to_collect = 20 #@param {type:"slider", min:5, max:50, step:1}
178
+ search_matching_type = "ratios" # @param ["ratios", "distances"]
179
+ distances_norm_order = 3 # @param {type:"slider", min:1, max:10, step:1}
180
+ maximum_match_ratio_to_search_for = 0.999 #@param {type:"slider", min:0, max:1, step:0.001}
181
+
182
+ print('=' * 70)
183
+ print('Master MIDI Dataset GPU Search and Filter')
184
+ print('=' * 70)
185
+
186
+ ###########
187
+
188
+ print('Loading MIDI files...')
189
+ print('This may take a while on a large dataset in particular.')
190
+
191
+ dataset_addr = "/content/Master-MIDI-Dataset"
192
+
193
+ filez = list()
194
+
195
+ for (dirpath, dirnames, filenames) in os.walk(dataset_addr):
196
+ for file in filenames:
197
+ if file.endswith(('.mid', '.midi', '.kar')):
198
+ filez.append(os.path.join(dirpath, file))
199
+
200
+ print('=' * 70)
201
+
202
+ if filez:
203
+
204
+ print('Randomizing file list...')
205
+ random.shuffle(filez)
206
+ print('=' * 70)
207
+
208
+ ###################
209
+
210
+ if not os.path.exists('/content/Output-MIDI-Dataset'):
211
+ os.makedirs('/content/Output-MIDI-Dataset')
212
+
213
+ ###################
214
+
215
+ input_files_count = 0
216
+ files_count = 0
217
+
218
+ for f in filez:
219
+ try:
220
+
221
+ input_files_count += 1
222
+
223
+ fn = os.path.basename(f)
224
+ fn1 = os.path.splitext(fn)[0]
225
+ ext = os.path.splitext(f)[1]
226
+
227
+ print('Processing MIDI File #', files_count+1, 'out of', len(filez))
228
+ print('MIDI file name', fn)
229
+ print('-' * 70)
230
+
231
+ #=======================================================
232
+
233
+ raw_score = TMIDIX.midi2single_track_ms_score(open(f, 'rb').read())
234
+ escore = TMIDIX.advanced_score_processor(raw_score, return_score_analysis=False, return_enhanced_score_notes=True)[0]
235
+
236
+ for e in escore:
237
+ e[1] = int(e[1] / 16)
238
+ e[2] = int(e[2] / 16)
239
+
240
+ src_sigs = []
241
+
242
+ for i in range(-6, 6):
243
+
244
+ escore_copy = copy.deepcopy(escore)
245
+
246
+ for e in escore_copy:
247
+ e[4] += i
248
+
249
+ cscore = TMIDIX.chordify_score([1000, escore_copy])
250
+
251
+ sig = []
252
+
253
+ for c in cscore:
254
+
255
+ pitches = sorted(set([p[4] for p in c if p[3] != 9]))
256
+
257
+ if pitches:
258
+ if len(pitches) > 1:
259
+ tones_chord = sorted(set([p % 12 for p in pitches]))
260
+ checked_tones_chord = TMIDIX.check_and_fix_tones_chord(tones_chord)
261
+
262
+ sig_token = TMIDIX.ALL_CHORDS.index(checked_tones_chord) + 128
263
+
264
+ elif len(pitches) == 1:
265
+ sig_token = pitches[0]
266
+
267
+ sig.append(sig_token)
268
+
269
+ fsig = [list(v) for v in Counter(sig).most_common()]
270
+
271
+ src_sig_mat = [0] * (len(TMIDIX.ALL_CHORDS)+128)
272
+
273
+ counts_sum = sum([c[1] for c in fsig])
274
+
275
+ for s in fsig:
276
+
277
+ src_sig_mat[s[0]] = s[1] / counts_sum
278
+
279
+ src_sigs.append(src_sig_mat)
280
+
281
+ src_signatures = cp.stack(cp.array(src_sigs))
282
+
283
+ #=======================================================
284
+
285
+ print('Searching for matches...Please wait...')
286
+ print('-' * 70)
287
+
288
+ lower_threshold = 0.0
289
+ upper_threshold = maximum_match_ratio_to_search_for
290
+ filter_size = number_of_top_matches_MIDIs_to_collect
291
+
292
+ final_ratios = []
293
+
294
+ avg_idxs = []
295
+
296
+ all_filtered_means = []
297
+ all_filtered_idxs = []
298
+ all_filtered_tvs = []
299
+
300
+ tv_idx = -6
301
+
302
+ for target_sig in tqdm(src_signatures):
303
+
304
+ if search_matching_type == 'ratios':
305
+
306
+ ratios = cp.where(target_sig != 0, cp.divide(cp.minimum(signatures_data, target_sig), cp.maximum(signatures_data, target_sig)), 0)
307
+ max_comp_lengths = cp.maximum(cp.repeat(cp.sum(target_sig != 0), signatures_data.shape[0]), cp.sum(signatures_data != 0, axis=1))
308
+
309
+ results = cp.divide(cp.sum(ratios, axis=1), max_comp_lengths)
310
+
311
+ elif search_matching_type == 'distances':
312
+
313
+ distances = cp.power(cp.sum(cp.power(cp.abs(signatures_data - target_sig), distances_norm_order), axis=1), 1 / distances_norm_order)
314
+
315
+ results = cp.max(distances) - distances
316
+
317
+ unique_means = cp.unique(results)
318
+ sorted_means = cp.sort(unique_means)[::-1]
319
+
320
+ filtered_means = sorted_means[(sorted_means >= lower_threshold) & (sorted_means <= upper_threshold)][:filter_size]
321
+
322
+ filtered_idxs = cp.where(cp.in1d(results, filtered_means))[0]
323
+
324
+ all_filtered_means.extend(results[cp.in1d(results, filtered_means)].tolist())
325
+
326
+ all_filtered_idxs.extend(filtered_idxs.tolist())
327
+
328
+ filtered_tvs = [tv_idx] * filtered_idxs.shape[0]
329
+
330
+ all_filtered_tvs.extend(filtered_tvs)
331
+
332
+ tv_idx += 1
333
+
334
+ filtered_results = sorted(zip(all_filtered_means, all_filtered_idxs, all_filtered_tvs), key=lambda x: x[0], reverse=True)[:filter_size]
335
+
336
+ #=======================================================
337
+
338
+ print('Done!')
339
+ print('-' * 70)
340
+ print('Max match ratio:', filtered_results[0][0])
341
+ print('Max match transpose value:', filtered_results[0][2])
342
+ print('Max match signature index:', filtered_results[0][1])
343
+ print('Max match file name:', signatures_file_names[filtered_results[0][1]])
344
+ print('-' * 70)
345
+ print('Copying max ratios MIDIs...')
346
+
347
+ for fr in filtered_results:
348
+
349
+ max_ratio_index = fr[1]
350
+
351
+ ffn = signatures_file_names[fr[1]]
352
+ ffn_idx = [y[0] for y in LAMD_files_list].index(ffn)
353
+
354
+ ff = LAMD_files_list[ffn_idx][1]
355
+
356
+ #=======================================================
357
+
358
+ dir_str = str(fn1)
359
+ copy_path = '/content/Output-MIDI-Dataset/'+dir_str
360
+ if not os.path.exists(copy_path):
361
+ os.mkdir(copy_path)
362
+
363
+ fff = str(fr[0] * 100) + '_' + str(fr[2]) + '_' + ffn + '.mid'
364
+
365
+ shutil.copy2(ff, copy_path+'/'+fff)
366
+
367
+ shutil.copy2(f, copy_path+'/'+fn)
368
+
369
+ #======================================================='''
370
+ print('Done!')
371
+ print('=' * 70)
372
+
373
+ #=======================================================
374
+
375
+ # Processed files counter
376
+ files_count += 1
377
+
378
+ except KeyboardInterrupt:
379
+ print('Quitting...')
380
+ print('Total number of processed MIDI files', files_count)
381
+ print('=' * 70)
382
+ break
383
+
384
+ except Exception as ex:
385
+ print('WARNING !!!')
386
+ print('=' * 70)
387
+ print('Bad file:', f)
388
+ print('Error detected:', ex)
389
+ print('=' * 70)
390
+ continue
391
+
392
+ print('Total number of processed MIDI files', files_count)
393
+ print('=' * 70)
394
+
395
+ else:
396
+ print('Could not find any MIDI files. Please check Dataset dir...')
397
+ print('=' * 70)
398
+
399
+ """# Congrats! You did it! :)"""