asigalov61
commited on
Commit
•
e2fb2a2
1
Parent(s):
bb5d53f
Upload 4 files
Browse files- MIDI_Images_Solo_Piano_Dataset_Maker.ipynb +492 -0
- TMIDIX.py +0 -0
- TPLOTS.py +1205 -0
- midi_images_solo_piano_dataset_maker.py +330 -0
MIDI_Images_Solo_Piano_Dataset_Maker.ipynb
ADDED
@@ -0,0 +1,492 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"nbformat": 4,
|
3 |
+
"nbformat_minor": 0,
|
4 |
+
"metadata": {
|
5 |
+
"colab": {
|
6 |
+
"private_outputs": true,
|
7 |
+
"provenance": []
|
8 |
+
},
|
9 |
+
"kernelspec": {
|
10 |
+
"name": "python3",
|
11 |
+
"display_name": "Python 3"
|
12 |
+
},
|
13 |
+
"language_info": {
|
14 |
+
"name": "python"
|
15 |
+
}
|
16 |
+
},
|
17 |
+
"cells": [
|
18 |
+
{
|
19 |
+
"cell_type": "markdown",
|
20 |
+
"source": [
|
21 |
+
"# MIDI Images Solo Piano Dataset Maker (ver. 1.0)\n",
|
22 |
+
"\n",
|
23 |
+
"***\n",
|
24 |
+
"\n",
|
25 |
+
"Powered by tegridy-tools: https://github.com/asigalov61/tegridy-tools\n",
|
26 |
+
"\n",
|
27 |
+
"***\n",
|
28 |
+
"\n",
|
29 |
+
"#### Project Los Angeles\n",
|
30 |
+
"\n",
|
31 |
+
"#### Tegridy Code 2024\n",
|
32 |
+
"\n",
|
33 |
+
"***"
|
34 |
+
],
|
35 |
+
"metadata": {
|
36 |
+
"id": "LUgrspEA-68o"
|
37 |
+
}
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"cell_type": "markdown",
|
41 |
+
"source": [
|
42 |
+
"# (SETUP ENVIRONMENT)"
|
43 |
+
],
|
44 |
+
"metadata": {
|
45 |
+
"id": "7N-KXNgQ_a0h"
|
46 |
+
}
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"cell_type": "code",
|
50 |
+
"execution_count": null,
|
51 |
+
"metadata": {
|
52 |
+
"id": "pxNxlyfZ8hCg",
|
53 |
+
"cellView": "form"
|
54 |
+
},
|
55 |
+
"outputs": [],
|
56 |
+
"source": [
|
57 |
+
"# @title Install dependecies\n",
|
58 |
+
"!git clone --depth 1 https://github.com/asigalov61/tegridy-tools"
|
59 |
+
]
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"cell_type": "code",
|
63 |
+
"source": [
|
64 |
+
"#@title Import all needed modules\n",
|
65 |
+
"\n",
|
66 |
+
"print('=' * 70)\n",
|
67 |
+
"print('Loading core modules...')\n",
|
68 |
+
"print('Please wait...')\n",
|
69 |
+
"print('=' * 70)\n",
|
70 |
+
"\n",
|
71 |
+
"import os\n",
|
72 |
+
"import copy\n",
|
73 |
+
"import math\n",
|
74 |
+
"import statistics\n",
|
75 |
+
"import random\n",
|
76 |
+
"import pickle\n",
|
77 |
+
"import shutil\n",
|
78 |
+
"from itertools import groupby\n",
|
79 |
+
"from collections import Counter\n",
|
80 |
+
"from sklearn.metrics import pairwise_distances\n",
|
81 |
+
"from sklearn import metrics\n",
|
82 |
+
"from joblib import Parallel, delayed, parallel_config\n",
|
83 |
+
"import numpy as np\n",
|
84 |
+
"from tqdm import tqdm\n",
|
85 |
+
"from PIL import Image\n",
|
86 |
+
"import matplotlib.pyplot as plt\n",
|
87 |
+
"\n",
|
88 |
+
"print('Done!')\n",
|
89 |
+
"print('=' * 70)\n",
|
90 |
+
"print('Creating I/O dirs...')\n",
|
91 |
+
"\n",
|
92 |
+
"if not os.path.exists('/content/Dataset'):\n",
|
93 |
+
" os.makedirs('/content/Dataset')\n",
|
94 |
+
"\n",
|
95 |
+
"print('Done!')\n",
|
96 |
+
"print('=' * 70)\n",
|
97 |
+
"print('Loading tegridy-tools modules...')\n",
|
98 |
+
"print('=' * 70)\n",
|
99 |
+
"\n",
|
100 |
+
"%cd /content/tegridy-tools/tegridy-tools\n",
|
101 |
+
"\n",
|
102 |
+
"import TMIDIX\n",
|
103 |
+
"import TMELODIES\n",
|
104 |
+
"import TPLOTS\n",
|
105 |
+
"import HaystackSearch\n",
|
106 |
+
"\n",
|
107 |
+
"%cd /content/\n",
|
108 |
+
"\n",
|
109 |
+
"print('=' * 70)\n",
|
110 |
+
"print('Done!')\n",
|
111 |
+
"print('=' * 70)"
|
112 |
+
],
|
113 |
+
"metadata": {
|
114 |
+
"id": "OblKfMMT8rfM",
|
115 |
+
"cellView": "form"
|
116 |
+
},
|
117 |
+
"execution_count": null,
|
118 |
+
"outputs": []
|
119 |
+
},
|
120 |
+
{
|
121 |
+
"cell_type": "markdown",
|
122 |
+
"source": [
|
123 |
+
"# (DOWNLOAD SAMPLE MIDI DATASET)"
|
124 |
+
],
|
125 |
+
"metadata": {
|
126 |
+
"id": "gUXM7WsN_ioe"
|
127 |
+
}
|
128 |
+
},
|
129 |
+
{
|
130 |
+
"cell_type": "code",
|
131 |
+
"source": [
|
132 |
+
"# @title Download sample MIDI dataset (POP909)\n",
|
133 |
+
"%cd /content/Dataset/\n",
|
134 |
+
"!git clone --depth 1 https://github.com/music-x-lab/POP909-Dataset\n",
|
135 |
+
"%cd /content/"
|
136 |
+
],
|
137 |
+
"metadata": {
|
138 |
+
"id": "JLm4OmOUYlEK",
|
139 |
+
"cellView": "form"
|
140 |
+
},
|
141 |
+
"execution_count": null,
|
142 |
+
"outputs": []
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"cell_type": "code",
|
146 |
+
"source": [
|
147 |
+
"#@title Save file list\n",
|
148 |
+
"###########\n",
|
149 |
+
"\n",
|
150 |
+
"print('=' * 70)\n",
|
151 |
+
"print('Loading MIDI files...')\n",
|
152 |
+
"print('This may take a while on a large dataset in particular...')\n",
|
153 |
+
"\n",
|
154 |
+
"dataset_addr = '/content/Dataset/'\n",
|
155 |
+
"\n",
|
156 |
+
"# os.chdir(dataset_addr)\n",
|
157 |
+
"filez = list()\n",
|
158 |
+
"for (dirpath, dirnames, filenames) in os.walk(dataset_addr):\n",
|
159 |
+
" filez += [os.path.join(dirpath, file) for file in filenames if file.endswith('.mid') or file.endswith('.midi') or file.endswith('.kar')]\n",
|
160 |
+
"print('=' * 70)\n",
|
161 |
+
"\n",
|
162 |
+
"if filez == []:\n",
|
163 |
+
" print('Could not find any MIDI files. Please check Dataset dir...')\n",
|
164 |
+
" print('=' * 70)\n",
|
165 |
+
"\n",
|
166 |
+
"print('Randomizing file list...')\n",
|
167 |
+
"random.shuffle(filez)\n",
|
168 |
+
"print('Done!')\n",
|
169 |
+
"print('=' * 70)\n",
|
170 |
+
"print('Total found MIDI files:', len(filez))\n",
|
171 |
+
"print('=' * 70)\n",
|
172 |
+
"\n",
|
173 |
+
"TMIDIX.Tegridy_Any_Pickle_File_Writer(filez, 'filez')\n",
|
174 |
+
"\n",
|
175 |
+
"print('=' * 70)"
|
176 |
+
],
|
177 |
+
"metadata": {
|
178 |
+
"cellView": "form",
|
179 |
+
"id": "AJrFrZ9grhMM"
|
180 |
+
},
|
181 |
+
"execution_count": null,
|
182 |
+
"outputs": []
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"cell_type": "markdown",
|
186 |
+
"source": [
|
187 |
+
"# (LOAD TMIDIX MIDI PROCESSOR)"
|
188 |
+
],
|
189 |
+
"metadata": {
|
190 |
+
"id": "RJeTdierAbeF"
|
191 |
+
}
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"cell_type": "code",
|
195 |
+
"source": [
|
196 |
+
"#@title Load TMIDIX MIDI processor\n",
|
197 |
+
"\n",
|
198 |
+
"print('=' * 70)\n",
|
199 |
+
"print('TMIDIX MIDI Processor')\n",
|
200 |
+
"print('=' * 70)\n",
|
201 |
+
"print('Loading...')\n",
|
202 |
+
"\n",
|
203 |
+
"###########\n",
|
204 |
+
"\n",
|
205 |
+
"def TMIDIX_MIDI_Processor(midi_file):\n",
|
206 |
+
"\n",
|
207 |
+
" fn = os.path.basename(midi_file)\n",
|
208 |
+
" fn1 = fn.split('.mid')[0]\n",
|
209 |
+
"\n",
|
210 |
+
" try:\n",
|
211 |
+
"\n",
|
212 |
+
" #=======================================================\n",
|
213 |
+
" # START PROCESSING\n",
|
214 |
+
"\n",
|
215 |
+
" raw_score = TMIDIX.midi2single_track_ms_score(midi_file)\n",
|
216 |
+
"\n",
|
217 |
+
" escore_notes = TMIDIX.advanced_score_processor(raw_score, return_enhanced_score_notes=True)[0]\n",
|
218 |
+
"\n",
|
219 |
+
" escore_notes = TMIDIX.augment_enhanced_score_notes(escore_notes, timings_divider=256)\n",
|
220 |
+
"\n",
|
221 |
+
" sp_escore_notes = TMIDIX.recalculate_score_timings(TMIDIX.solo_piano_escore_notes(escore_notes, keep_drums=False))\n",
|
222 |
+
"\n",
|
223 |
+
" if sp_escore_notes:\n",
|
224 |
+
"\n",
|
225 |
+
" bmatrix = TMIDIX.escore_notes_to_binary_matrix(sp_escore_notes)\n",
|
226 |
+
"\n",
|
227 |
+
" return [fn1, bmatrix]\n",
|
228 |
+
"\n",
|
229 |
+
" else:\n",
|
230 |
+
" return [fn1, []]\n",
|
231 |
+
"\n",
|
232 |
+
" #=======================================================\n",
|
233 |
+
"\n",
|
234 |
+
" except Exception as ex:\n",
|
235 |
+
" print('WARNING !!!')\n",
|
236 |
+
" print('=' * 70)\n",
|
237 |
+
" print('Bad MIDI:', midi_file)\n",
|
238 |
+
" print('Error detected:', ex)\n",
|
239 |
+
" print('=' * 70)\n",
|
240 |
+
" return None\n",
|
241 |
+
"\n",
|
242 |
+
"print('Done!')\n",
|
243 |
+
"print('=' * 70)"
|
244 |
+
],
|
245 |
+
"metadata": {
|
246 |
+
"cellView": "form",
|
247 |
+
"id": "fBbIiUWSZA5y"
|
248 |
+
},
|
249 |
+
"execution_count": null,
|
250 |
+
"outputs": []
|
251 |
+
},
|
252 |
+
{
|
253 |
+
"cell_type": "markdown",
|
254 |
+
"source": [
|
255 |
+
"# (PROCESS MIDIs)"
|
256 |
+
],
|
257 |
+
"metadata": {
|
258 |
+
"id": "R3QxQN6OA_jX"
|
259 |
+
}
|
260 |
+
},
|
261 |
+
{
|
262 |
+
"cell_type": "code",
|
263 |
+
"source": [
|
264 |
+
"#@title Process MIDIs with TMIDIX MIDI processor\n",
|
265 |
+
"output_folder = \"/content/MIDI-Images/\" # @param {\"type\":\"string\"}\n",
|
266 |
+
"\n",
|
267 |
+
"NUMBER_OF_PARALLEL_JOBS = 4 # Number of parallel jobs\n",
|
268 |
+
"NUMBER_OF_FILES_PER_ITERATION = 4 # Number of files to queue for each parallel iteration\n",
|
269 |
+
"SAVE_EVERY_NUMBER_OF_ITERATIONS = 128 # Save every 2560 files\n",
|
270 |
+
"\n",
|
271 |
+
"print('=' * 70)\n",
|
272 |
+
"print('TMIDIX MIDI Processor')\n",
|
273 |
+
"print('=' * 70)\n",
|
274 |
+
"print('Starting up...')\n",
|
275 |
+
"print('=' * 70)\n",
|
276 |
+
"\n",
|
277 |
+
"###########\n",
|
278 |
+
"\n",
|
279 |
+
"melody_chords_f = []\n",
|
280 |
+
"\n",
|
281 |
+
"files_count = 0\n",
|
282 |
+
"\n",
|
283 |
+
"print('Processing MIDI files...')\n",
|
284 |
+
"print('Please wait...')\n",
|
285 |
+
"print('=' * 70)\n",
|
286 |
+
"\n",
|
287 |
+
"for i in tqdm(range(0, len(filez), NUMBER_OF_FILES_PER_ITERATION)):\n",
|
288 |
+
"\n",
|
289 |
+
" with parallel_config(backend='threading', n_jobs=NUMBER_OF_PARALLEL_JOBS, verbose = 0):\n",
|
290 |
+
"\n",
|
291 |
+
" output = Parallel(n_jobs=NUMBER_OF_PARALLEL_JOBS, verbose=0)(delayed(TMIDIX_MIDI_Processor)(f) for f in filez[i:i+NUMBER_OF_FILES_PER_ITERATION])\n",
|
292 |
+
"\n",
|
293 |
+
" for o in output:\n",
|
294 |
+
"\n",
|
295 |
+
" if o is not None:\n",
|
296 |
+
" melody_chords_f.append(o)\n",
|
297 |
+
"\n",
|
298 |
+
" if i % (NUMBER_OF_FILES_PER_ITERATION * SAVE_EVERY_NUMBER_OF_ITERATIONS) == 0 and i != 0:\n",
|
299 |
+
"\n",
|
300 |
+
" print('SAVING !!!')\n",
|
301 |
+
" print('=' * 70)\n",
|
302 |
+
" print('Saving processed files...')\n",
|
303 |
+
" files_count += len(melody_chords_f)\n",
|
304 |
+
" print('=' * 70)\n",
|
305 |
+
" print('Processed so far:', files_count, 'out of', len(filez), '===', files_count / len(filez), 'good files ratio')\n",
|
306 |
+
" print('=' * 70)\n",
|
307 |
+
" print('Writing images...')\n",
|
308 |
+
" print('Please wait...')\n",
|
309 |
+
"\n",
|
310 |
+
" for mat in melody_chords_f:\n",
|
311 |
+
"\n",
|
312 |
+
" if mat[1]:\n",
|
313 |
+
"\n",
|
314 |
+
" TPLOTS.binary_matrix_to_images(mat[1],\n",
|
315 |
+
" 128,\n",
|
316 |
+
" 32,\n",
|
317 |
+
" output_folder=output_folder+str(mat[0])+'/',\n",
|
318 |
+
" output_img_prefix=str(mat[0]),\n",
|
319 |
+
" output_img_ext='.png',\n",
|
320 |
+
" verbose=False\n",
|
321 |
+
" )\n",
|
322 |
+
"\n",
|
323 |
+
" print('Done!')\n",
|
324 |
+
" print('=' * 70)\n",
|
325 |
+
" melody_chords_f = []\n",
|
326 |
+
"\n",
|
327 |
+
"print('SAVING !!!')\n",
|
328 |
+
"print('=' * 70)\n",
|
329 |
+
"print('Saving processed files...')\n",
|
330 |
+
"files_count += len(melody_chords_f)\n",
|
331 |
+
"print('=' * 70)\n",
|
332 |
+
"print('Processed so far:', files_count, 'out of', len(filez), '===', files_count / len(filez), 'good files ratio')\n",
|
333 |
+
"print('=' * 70)\n",
|
334 |
+
"print('Writing images...')\n",
|
335 |
+
"print('Please wait...')\n",
|
336 |
+
"\n",
|
337 |
+
"for mat in melody_chords_f:\n",
|
338 |
+
"\n",
|
339 |
+
" if mat[1]:\n",
|
340 |
+
"\n",
|
341 |
+
" TPLOTS.binary_matrix_to_images(mat[1],\n",
|
342 |
+
" 128,\n",
|
343 |
+
" 32,\n",
|
344 |
+
" output_folder=output_folder+str(mat[0])+'/',\n",
|
345 |
+
" output_img_prefix=str(mat[0]),\n",
|
346 |
+
" output_img_ext='.png',\n",
|
347 |
+
" verbose=False\n",
|
348 |
+
" )\n",
|
349 |
+
"\n",
|
350 |
+
"print('Done!')\n",
|
351 |
+
"print('=' * 70)"
|
352 |
+
],
|
353 |
+
"metadata": {
|
354 |
+
"cellView": "form",
|
355 |
+
"id": "15y4uzSOZX52"
|
356 |
+
},
|
357 |
+
"execution_count": null,
|
358 |
+
"outputs": []
|
359 |
+
},
|
360 |
+
{
|
361 |
+
"cell_type": "markdown",
|
362 |
+
"source": [
|
363 |
+
"# (LOAD IMAGES)"
|
364 |
+
],
|
365 |
+
"metadata": {
|
366 |
+
"id": "GtejvUFAFocZ"
|
367 |
+
}
|
368 |
+
},
|
369 |
+
{
|
370 |
+
"cell_type": "code",
|
371 |
+
"source": [
|
372 |
+
"#@title Load created MIDI images\n",
|
373 |
+
"full_path_to_metadata_pickle_files = \"/content/MIDI-Images\" #@param {type:\"string\"}\n",
|
374 |
+
"\n",
|
375 |
+
"print('=' * 70)\n",
|
376 |
+
"print('MIDI Images Reader')\n",
|
377 |
+
"print('=' * 70)\n",
|
378 |
+
"print('Searching for images...')\n",
|
379 |
+
"\n",
|
380 |
+
"filez = list()\n",
|
381 |
+
"for (dirpath, dirnames, filenames) in os.walk(full_path_to_metadata_pickle_files):\n",
|
382 |
+
" filez += [os.path.join(dirpath, file) for file in filenames if file.endswith('.png')]\n",
|
383 |
+
"print('=' * 70)\n",
|
384 |
+
"\n",
|
385 |
+
"filez.sort()\n",
|
386 |
+
"\n",
|
387 |
+
"print('Found', len(filez), 'images!')\n",
|
388 |
+
"print('=' * 70)\n",
|
389 |
+
"print('Reading images...')\n",
|
390 |
+
"print('Please wait...')\n",
|
391 |
+
"print('=' * 70)\n",
|
392 |
+
"\n",
|
393 |
+
"fidx = 0\n",
|
394 |
+
"\n",
|
395 |
+
"all_read_images = []\n",
|
396 |
+
"\n",
|
397 |
+
"for img in tqdm(filez):\n",
|
398 |
+
"\n",
|
399 |
+
" img = Image.open(img)\n",
|
400 |
+
"\n",
|
401 |
+
" img_arr = np.array(img).tolist()\n",
|
402 |
+
"\n",
|
403 |
+
" all_read_images.append(img_arr)\n",
|
404 |
+
"\n",
|
405 |
+
" fidx += 1\n",
|
406 |
+
"\n",
|
407 |
+
"print('Done!')\n",
|
408 |
+
"print('=' * 70)\n",
|
409 |
+
"print('Loaded', fidx, 'images!')\n",
|
410 |
+
"print('=' * 70)\n",
|
411 |
+
"print('Done!')\n",
|
412 |
+
"print('=' * 70)"
|
413 |
+
],
|
414 |
+
"metadata": {
|
415 |
+
"cellView": "form",
|
416 |
+
"id": "cXpLWHG1dBB3"
|
417 |
+
},
|
418 |
+
"execution_count": null,
|
419 |
+
"outputs": []
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"cell_type": "markdown",
|
423 |
+
"source": [
|
424 |
+
"# (TEST IMAGES)"
|
425 |
+
],
|
426 |
+
"metadata": {
|
427 |
+
"id": "qbClHSmhB1NF"
|
428 |
+
}
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"cell_type": "code",
|
432 |
+
"source": [
|
433 |
+
"# @title Test created MIDI images\n",
|
434 |
+
"\n",
|
435 |
+
"print('=' * 70)\n",
|
436 |
+
"\n",
|
437 |
+
"image = random.choice(all_read_images)\n",
|
438 |
+
"\n",
|
439 |
+
"escore = TMIDIX.binary_matrix_to_original_escore_notes(image)\n",
|
440 |
+
"\n",
|
441 |
+
"output_score, patches, overflow_patches = TMIDIX.patch_enhanced_score_notes(escore)\n",
|
442 |
+
"\n",
|
443 |
+
"detailed_stats = TMIDIX.Tegridy_ms_SONG_to_MIDI_Converter(output_score,\n",
|
444 |
+
" output_signature = 'MIDI Images',\n",
|
445 |
+
" output_file_name = '/content/MIDI-Images-Composition',\n",
|
446 |
+
" track_name='Project Los Angeles',\n",
|
447 |
+
" list_of_MIDI_patches=patches,\n",
|
448 |
+
" timings_multiplier=256\n",
|
449 |
+
" )\n",
|
450 |
+
"\n",
|
451 |
+
"print('=' * 70)"
|
452 |
+
],
|
453 |
+
"metadata": {
|
454 |
+
"id": "nrPDM1VQdKES",
|
455 |
+
"cellView": "form"
|
456 |
+
},
|
457 |
+
"execution_count": null,
|
458 |
+
"outputs": []
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"cell_type": "markdown",
|
462 |
+
"source": [
|
463 |
+
"# (ZIP IMAGES)"
|
464 |
+
],
|
465 |
+
"metadata": {
|
466 |
+
"id": "sIq55gvPCgJh"
|
467 |
+
}
|
468 |
+
},
|
469 |
+
{
|
470 |
+
"cell_type": "code",
|
471 |
+
"source": [
|
472 |
+
"# @title Zip created MIDI images\n",
|
473 |
+
"!zip -9 -r POP909_MIDI_Images_128_128_32_BW.zip MIDI-Images/ > /dev/null"
|
474 |
+
],
|
475 |
+
"metadata": {
|
476 |
+
"id": "tVe0REKSqJeV",
|
477 |
+
"cellView": "form"
|
478 |
+
},
|
479 |
+
"execution_count": null,
|
480 |
+
"outputs": []
|
481 |
+
},
|
482 |
+
{
|
483 |
+
"cell_type": "markdown",
|
484 |
+
"source": [
|
485 |
+
"# Congrats! You did it! :)"
|
486 |
+
],
|
487 |
+
"metadata": {
|
488 |
+
"id": "iDdMYg4haGFn"
|
489 |
+
}
|
490 |
+
}
|
491 |
+
]
|
492 |
+
}
|
TMIDIX.py
ADDED
The diff for this file is too large to render.
See raw diff
|
|
TPLOTS.py
ADDED
@@ -0,0 +1,1205 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#! /usr/bin/python3
|
2 |
+
|
3 |
+
r'''############################################################################
|
4 |
+
################################################################################
|
5 |
+
#
|
6 |
+
#
|
7 |
+
# Tegridy Plots Python Module (TPLOTS)
|
8 |
+
# Version 1.0
|
9 |
+
#
|
10 |
+
# Project Los Angeles
|
11 |
+
#
|
12 |
+
# Tegridy Code 2024
|
13 |
+
#
|
14 |
+
# https://github.com/asigalov61/tegridy-tools
|
15 |
+
#
|
16 |
+
#
|
17 |
+
################################################################################
|
18 |
+
#
|
19 |
+
# Copyright 2024 Project Los Angeles / Tegridy Code
|
20 |
+
#
|
21 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
22 |
+
# you may not use this file except in compliance with the License.
|
23 |
+
# You may obtain a copy of the License at
|
24 |
+
#
|
25 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
26 |
+
#
|
27 |
+
# Unless required by applicable law or agreed to in writing, software
|
28 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
29 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
30 |
+
# See the License for the specific language governing permissions and
|
31 |
+
# limitations under the License.
|
32 |
+
#
|
33 |
+
################################################################################
|
34 |
+
################################################################################
|
35 |
+
#
|
36 |
+
# Critical dependencies
|
37 |
+
#
|
38 |
+
# !pip install numpy
|
39 |
+
# !pip install scipy
|
40 |
+
# !pip install matplotlib
|
41 |
+
# !pip install networkx
|
42 |
+
# !pip3 install scikit-learn
|
43 |
+
#
|
44 |
+
################################################################################
|
45 |
+
#
|
46 |
+
# Future critical dependencies
|
47 |
+
#
|
48 |
+
# !pip install umap-learn
|
49 |
+
# !pip install alphashape
|
50 |
+
#
|
51 |
+
################################################################################
|
52 |
+
'''
|
53 |
+
|
54 |
+
################################################################################
|
55 |
+
# Modules imports
|
56 |
+
################################################################################
|
57 |
+
|
58 |
+
import os
|
59 |
+
from collections import Counter
|
60 |
+
from itertools import groupby
|
61 |
+
|
62 |
+
import numpy as np
|
63 |
+
|
64 |
+
import networkx as nx
|
65 |
+
|
66 |
+
from sklearn.manifold import TSNE
|
67 |
+
from sklearn import metrics
|
68 |
+
from sklearn.preprocessing import MinMaxScaler
|
69 |
+
from sklearn.decomposition import PCA
|
70 |
+
|
71 |
+
from scipy.ndimage import zoom
|
72 |
+
from scipy.spatial import distance_matrix
|
73 |
+
from scipy.sparse.csgraph import minimum_spanning_tree
|
74 |
+
from scipy.stats import zscore
|
75 |
+
|
76 |
+
import matplotlib.pyplot as plt
|
77 |
+
from PIL import Image
|
78 |
+
|
79 |
+
################################################################################
|
80 |
+
# Constants
|
81 |
+
################################################################################
|
82 |
+
|
83 |
+
ALL_CHORDS_FILTERED = [[0], [0, 3], [0, 3, 5], [0, 3, 5, 8], [0, 3, 5, 9], [0, 3, 5, 10], [0, 3, 7],
|
84 |
+
[0, 3, 7, 10], [0, 3, 8], [0, 3, 9], [0, 3, 10], [0, 4], [0, 4, 6],
|
85 |
+
[0, 4, 6, 9], [0, 4, 6, 10], [0, 4, 7], [0, 4, 7, 10], [0, 4, 8], [0, 4, 9],
|
86 |
+
[0, 4, 10], [0, 5], [0, 5, 8], [0, 5, 9], [0, 5, 10], [0, 6], [0, 6, 9],
|
87 |
+
[0, 6, 10], [0, 7], [0, 7, 10], [0, 8], [0, 9], [0, 10], [1], [1, 4],
|
88 |
+
[1, 4, 6], [1, 4, 6, 9], [1, 4, 6, 10], [1, 4, 6, 11], [1, 4, 7],
|
89 |
+
[1, 4, 7, 10], [1, 4, 7, 11], [1, 4, 8], [1, 4, 8, 11], [1, 4, 9], [1, 4, 10],
|
90 |
+
[1, 4, 11], [1, 5], [1, 5, 8], [1, 5, 8, 11], [1, 5, 9], [1, 5, 10],
|
91 |
+
[1, 5, 11], [1, 6], [1, 6, 9], [1, 6, 10], [1, 6, 11], [1, 7], [1, 7, 10],
|
92 |
+
[1, 7, 11], [1, 8], [1, 8, 11], [1, 9], [1, 10], [1, 11], [2], [2, 5],
|
93 |
+
[2, 5, 8], [2, 5, 8, 11], [2, 5, 9], [2, 5, 10], [2, 5, 11], [2, 6], [2, 6, 9],
|
94 |
+
[2, 6, 10], [2, 6, 11], [2, 7], [2, 7, 10], [2, 7, 11], [2, 8], [2, 8, 11],
|
95 |
+
[2, 9], [2, 10], [2, 11], [3], [3, 5], [3, 5, 8], [3, 5, 8, 11], [3, 5, 9],
|
96 |
+
[3, 5, 10], [3, 5, 11], [3, 7], [3, 7, 10], [3, 7, 11], [3, 8], [3, 8, 11],
|
97 |
+
[3, 9], [3, 10], [3, 11], [4], [4, 6], [4, 6, 9], [4, 6, 10], [4, 6, 11],
|
98 |
+
[4, 7], [4, 7, 10], [4, 7, 11], [4, 8], [4, 8, 11], [4, 9], [4, 10], [4, 11],
|
99 |
+
[5], [5, 8], [5, 8, 11], [5, 9], [5, 10], [5, 11], [6], [6, 9], [6, 10],
|
100 |
+
[6, 11], [7], [7, 10], [7, 11], [8], [8, 11], [9], [10], [11]]
|
101 |
+
|
102 |
+
################################################################################
|
103 |
+
|
104 |
+
CHORDS_TYPES = ['WHITE', 'BLACK', 'UNKNOWN', 'MIXED WHITE', 'MIXED BLACK', 'MIXED GRAY']
|
105 |
+
|
106 |
+
################################################################################
|
107 |
+
|
108 |
+
WHITE_NOTES = [0, 2, 4, 5, 7, 9, 11]
|
109 |
+
|
110 |
+
################################################################################
|
111 |
+
|
112 |
+
BLACK_NOTES = [1, 3, 6, 8, 10]
|
113 |
+
|
114 |
+
################################################################################
|
115 |
+
# Helper functions
|
116 |
+
################################################################################
|
117 |
+
|
118 |
+
def tones_chord_type(tones_chord,
|
119 |
+
return_chord_type_index=True,
|
120 |
+
):
|
121 |
+
|
122 |
+
"""
|
123 |
+
Returns tones chord type
|
124 |
+
"""
|
125 |
+
|
126 |
+
WN = WHITE_NOTES
|
127 |
+
BN = BLACK_NOTES
|
128 |
+
MX = WHITE_NOTES + BLACK_NOTES
|
129 |
+
|
130 |
+
|
131 |
+
CHORDS = ALL_CHORDS_FILTERED
|
132 |
+
|
133 |
+
tones_chord = sorted(tones_chord)
|
134 |
+
|
135 |
+
ctype = 'UNKNOWN'
|
136 |
+
|
137 |
+
if tones_chord in CHORDS:
|
138 |
+
|
139 |
+
if sorted(set(tones_chord) & set(WN)) == tones_chord:
|
140 |
+
ctype = 'WHITE'
|
141 |
+
|
142 |
+
elif sorted(set(tones_chord) & set(BN)) == tones_chord:
|
143 |
+
ctype = 'BLACK'
|
144 |
+
|
145 |
+
if len(tones_chord) > 1 and sorted(set(tones_chord) & set(MX)) == tones_chord:
|
146 |
+
|
147 |
+
if len(sorted(set(tones_chord) & set(WN))) == len(sorted(set(tones_chord) & set(BN))):
|
148 |
+
ctype = 'MIXED GRAY'
|
149 |
+
|
150 |
+
elif len(sorted(set(tones_chord) & set(WN))) > len(sorted(set(tones_chord) & set(BN))):
|
151 |
+
ctype = 'MIXED WHITE'
|
152 |
+
|
153 |
+
elif len(sorted(set(tones_chord) & set(WN))) < len(sorted(set(tones_chord) & set(BN))):
|
154 |
+
ctype = 'MIXED BLACK'
|
155 |
+
|
156 |
+
if return_chord_type_index:
|
157 |
+
return CHORDS_TYPES.index(ctype)
|
158 |
+
|
159 |
+
else:
|
160 |
+
return ctype
|
161 |
+
|
162 |
+
###################################################################################
|
163 |
+
|
164 |
+
def tone_type(tone,
|
165 |
+
return_tone_type_index=True
|
166 |
+
):
|
167 |
+
|
168 |
+
"""
|
169 |
+
Returns tone type
|
170 |
+
"""
|
171 |
+
|
172 |
+
tone = tone % 12
|
173 |
+
|
174 |
+
if tone in BLACK_NOTES:
|
175 |
+
if return_tone_type_index:
|
176 |
+
return CHORDS_TYPES.index('BLACK')
|
177 |
+
else:
|
178 |
+
return "BLACK"
|
179 |
+
|
180 |
+
else:
|
181 |
+
if return_tone_type_index:
|
182 |
+
return CHORDS_TYPES.index('WHITE')
|
183 |
+
else:
|
184 |
+
return "WHITE"
|
185 |
+
|
186 |
+
###################################################################################
|
187 |
+
|
188 |
+
def find_closest_points(points, return_points=True):
|
189 |
+
|
190 |
+
"""
|
191 |
+
Find closest 2D points
|
192 |
+
"""
|
193 |
+
|
194 |
+
coords = np.array(points)
|
195 |
+
|
196 |
+
num_points = coords.shape[0]
|
197 |
+
closest_matches = np.zeros(num_points, dtype=int)
|
198 |
+
distances = np.zeros((num_points, num_points))
|
199 |
+
|
200 |
+
for i in range(num_points):
|
201 |
+
for j in range(num_points):
|
202 |
+
if i != j:
|
203 |
+
distances[i, j] = np.linalg.norm(coords[i] - coords[j])
|
204 |
+
else:
|
205 |
+
distances[i, j] = np.inf
|
206 |
+
|
207 |
+
closest_matches = np.argmin(distances, axis=1)
|
208 |
+
|
209 |
+
if return_points:
|
210 |
+
points_matches = coords[closest_matches].tolist()
|
211 |
+
return points_matches
|
212 |
+
|
213 |
+
else:
|
214 |
+
return closest_matches.tolist()
|
215 |
+
|
216 |
+
################################################################################
|
217 |
+
|
218 |
+
def reduce_dimensionality_tsne(list_of_valies,
|
219 |
+
n_comp=2,
|
220 |
+
n_iter=5000,
|
221 |
+
verbose=True
|
222 |
+
):
|
223 |
+
|
224 |
+
"""
|
225 |
+
Reduces the dimensionality of the values using t-SNE.
|
226 |
+
"""
|
227 |
+
|
228 |
+
vals = np.array(list_of_valies)
|
229 |
+
|
230 |
+
tsne = TSNE(n_components=n_comp,
|
231 |
+
n_iter=n_iter,
|
232 |
+
verbose=verbose)
|
233 |
+
|
234 |
+
reduced_vals = tsne.fit_transform(vals)
|
235 |
+
|
236 |
+
return reduced_vals.tolist()
|
237 |
+
|
238 |
+
################################################################################
|
239 |
+
|
240 |
+
def compute_mst_edges(similarity_scores_list):
|
241 |
+
|
242 |
+
"""
|
243 |
+
Computes the Minimum Spanning Tree (MST) edges based on the similarity scores.
|
244 |
+
"""
|
245 |
+
|
246 |
+
num_tokens = len(similarity_scores_list[0])
|
247 |
+
|
248 |
+
graph = nx.Graph()
|
249 |
+
|
250 |
+
for i in range(num_tokens):
|
251 |
+
for j in range(i + 1, num_tokens):
|
252 |
+
weight = 1 - similarity_scores_list[i][j]
|
253 |
+
graph.add_edge(i, j, weight=weight)
|
254 |
+
|
255 |
+
mst = nx.minimum_spanning_tree(graph)
|
256 |
+
|
257 |
+
mst_edges = list(mst.edges(data=False))
|
258 |
+
|
259 |
+
return mst_edges
|
260 |
+
|
261 |
+
################################################################################
|
262 |
+
|
263 |
+
def square_binary_matrix(binary_matrix,
|
264 |
+
matrix_size=128,
|
265 |
+
interpolation_order=5,
|
266 |
+
return_square_matrix_points=False
|
267 |
+
):
|
268 |
+
|
269 |
+
"""
|
270 |
+
Reduces an arbitrary binary matrix to a square binary matrix
|
271 |
+
"""
|
272 |
+
|
273 |
+
zoom_factors = (matrix_size / len(binary_matrix), 1)
|
274 |
+
|
275 |
+
resized_matrix = zoom(binary_matrix, zoom_factors, order=interpolation_order)
|
276 |
+
|
277 |
+
resized_matrix = (resized_matrix > 0.5).astype(int)
|
278 |
+
|
279 |
+
final_matrix = np.zeros((matrix_size, matrix_size), dtype=int)
|
280 |
+
final_matrix[:, :resized_matrix.shape[1]] = resized_matrix
|
281 |
+
|
282 |
+
points = np.column_stack(np.where(final_matrix == 1)).tolist()
|
283 |
+
|
284 |
+
if return_square_matrix_points:
|
285 |
+
return points
|
286 |
+
|
287 |
+
else:
|
288 |
+
return resized_matrix
|
289 |
+
|
290 |
+
################################################################################
|
291 |
+
|
292 |
+
def square_matrix_points_colors(square_matrix_points):
|
293 |
+
|
294 |
+
"""
|
295 |
+
Returns colors for square matrix points
|
296 |
+
"""
|
297 |
+
|
298 |
+
cmap = generate_colors(12)
|
299 |
+
|
300 |
+
chords = []
|
301 |
+
chords_dict = set()
|
302 |
+
counts = []
|
303 |
+
|
304 |
+
for k, v in groupby(square_matrix_points, key=lambda x: x[0]):
|
305 |
+
pgroup = [vv[1] for vv in v]
|
306 |
+
chord = sorted(set(pgroup))
|
307 |
+
tchord = sorted(set([p % 12 for p in chord]))
|
308 |
+
chords_dict.add(tuple(tchord))
|
309 |
+
chords.append(tuple(tchord))
|
310 |
+
counts.append(len(pgroup))
|
311 |
+
|
312 |
+
chords_dict = sorted(chords_dict)
|
313 |
+
|
314 |
+
colors = []
|
315 |
+
|
316 |
+
for i, c in enumerate(chords):
|
317 |
+
colors.extend([cmap[round(sum(c) / len(c))]] * counts[i])
|
318 |
+
|
319 |
+
return colors
|
320 |
+
|
321 |
+
################################################################################
|
322 |
+
|
323 |
+
def hsv_to_rgb(h, s, v):
|
324 |
+
|
325 |
+
if s == 0.0:
|
326 |
+
return v, v, v
|
327 |
+
|
328 |
+
i = int(h*6.0)
|
329 |
+
f = (h*6.0) - i
|
330 |
+
p = v*(1.0 - s)
|
331 |
+
q = v*(1.0 - s*f)
|
332 |
+
t = v*(1.0 - s*(1.0-f))
|
333 |
+
i = i%6
|
334 |
+
|
335 |
+
return [(v, t, p), (q, v, p), (p, v, t), (p, q, v), (t, p, v), (v, p, q)][i]
|
336 |
+
|
337 |
+
################################################################################
|
338 |
+
|
339 |
+
def generate_colors(n):
|
340 |
+
return [hsv_to_rgb(i/n, 1, 1) for i in range(n)]
|
341 |
+
|
342 |
+
################################################################################
|
343 |
+
|
344 |
+
def add_arrays(a, b):
|
345 |
+
return [sum(pair) for pair in zip(a, b)]
|
346 |
+
|
347 |
+
################################################################################
|
348 |
+
|
349 |
+
def calculate_similarities(lists_of_values, metric='cosine'):
|
350 |
+
return metrics.pairwise_distances(lists_of_values, metric=metric).tolist()
|
351 |
+
|
352 |
+
################################################################################
|
353 |
+
|
354 |
+
def get_tokens_embeddings(x_transformer_model):
|
355 |
+
return x_transformer_model.net.token_emb.emb.weight.detach().cpu().tolist()
|
356 |
+
|
357 |
+
################################################################################
|
358 |
+
|
359 |
+
def minkowski_distance_matrix(X, p=3):
|
360 |
+
|
361 |
+
X = np.array(X)
|
362 |
+
|
363 |
+
n = X.shape[0]
|
364 |
+
dist_matrix = np.zeros((n, n))
|
365 |
+
|
366 |
+
for i in range(n):
|
367 |
+
for j in range(n):
|
368 |
+
dist_matrix[i, j] = np.sum(np.abs(X[i] - X[j])**p)**(1/p)
|
369 |
+
|
370 |
+
return dist_matrix.tolist()
|
371 |
+
|
372 |
+
################################################################################
|
373 |
+
|
374 |
+
def robust_normalize(values):
|
375 |
+
|
376 |
+
values = np.array(values)
|
377 |
+
q1 = np.percentile(values, 25)
|
378 |
+
q3 = np.percentile(values, 75)
|
379 |
+
iqr = q3 - q1
|
380 |
+
|
381 |
+
filtered_values = values[(values >= q1 - 1.5 * iqr) & (values <= q3 + 1.5 * iqr)]
|
382 |
+
|
383 |
+
min_val = np.min(filtered_values)
|
384 |
+
max_val = np.max(filtered_values)
|
385 |
+
normalized_values = (values - min_val) / (max_val - min_val)
|
386 |
+
|
387 |
+
normalized_values = np.clip(normalized_values, 0, 1)
|
388 |
+
|
389 |
+
return normalized_values.tolist()
|
390 |
+
|
391 |
+
################################################################################
|
392 |
+
|
393 |
+
def min_max_normalize(values):
|
394 |
+
|
395 |
+
scaler = MinMaxScaler()
|
396 |
+
|
397 |
+
return scaler.fit_transform(values).tolist()
|
398 |
+
|
399 |
+
################################################################################
|
400 |
+
|
401 |
+
def remove_points_outliers(points, z_score_threshold=3):
|
402 |
+
|
403 |
+
points = np.array(points)
|
404 |
+
|
405 |
+
z_scores = np.abs(zscore(points, axis=0))
|
406 |
+
|
407 |
+
return points[(z_scores < z_score_threshold).all(axis=1)].tolist()
|
408 |
+
|
409 |
+
################################################################################
|
410 |
+
|
411 |
+
def generate_labels(lists_of_values,
|
412 |
+
return_indices_labels=False
|
413 |
+
):
|
414 |
+
|
415 |
+
ordered_indices = list(range(len(lists_of_values)))
|
416 |
+
ordered_indices_labels = [str(i) for i in ordered_indices]
|
417 |
+
ordered_values_labels = [str(lists_of_values[i]) for i in ordered_indices]
|
418 |
+
|
419 |
+
if return_indices_labels:
|
420 |
+
return ordered_indices_labels
|
421 |
+
|
422 |
+
else:
|
423 |
+
return ordered_values_labels
|
424 |
+
|
425 |
+
################################################################################
|
426 |
+
|
427 |
+
def reduce_dimensionality_pca(list_of_values, n_components=2):
|
428 |
+
|
429 |
+
"""
|
430 |
+
Reduces the dimensionality of the values using PCA.
|
431 |
+
"""
|
432 |
+
|
433 |
+
pca = PCA(n_components=n_components)
|
434 |
+
pca_data = pca.fit_transform(list_of_values)
|
435 |
+
|
436 |
+
return pca_data.tolist()
|
437 |
+
|
438 |
+
def reduce_dimensionality_simple(list_of_values,
|
439 |
+
return_means=True,
|
440 |
+
return_std_devs=True,
|
441 |
+
return_medians=False,
|
442 |
+
return_vars=False
|
443 |
+
):
|
444 |
+
|
445 |
+
'''
|
446 |
+
Reduces dimensionality of the values in a simple way
|
447 |
+
'''
|
448 |
+
|
449 |
+
array = np.array(list_of_values)
|
450 |
+
results = []
|
451 |
+
|
452 |
+
if return_means:
|
453 |
+
means = np.mean(array, axis=1)
|
454 |
+
results.append(means)
|
455 |
+
|
456 |
+
if return_std_devs:
|
457 |
+
std_devs = np.std(array, axis=1)
|
458 |
+
results.append(std_devs)
|
459 |
+
|
460 |
+
if return_medians:
|
461 |
+
medians = np.median(array, axis=1)
|
462 |
+
results.append(medians)
|
463 |
+
|
464 |
+
if return_vars:
|
465 |
+
vars = np.var(array, axis=1)
|
466 |
+
results.append(vars)
|
467 |
+
|
468 |
+
merged_results = np.column_stack(results)
|
469 |
+
|
470 |
+
return merged_results.tolist()
|
471 |
+
|
472 |
+
################################################################################
|
473 |
+
|
474 |
+
def reduce_dimensionality_2d_distance(list_of_values, p=5):
|
475 |
+
|
476 |
+
'''
|
477 |
+
Reduces the dimensionality of the values using 2d distance
|
478 |
+
'''
|
479 |
+
|
480 |
+
values = np.array(list_of_values)
|
481 |
+
|
482 |
+
dist_matrix = distance_matrix(values, values, p=p)
|
483 |
+
|
484 |
+
mst = minimum_spanning_tree(dist_matrix).toarray()
|
485 |
+
|
486 |
+
points = []
|
487 |
+
|
488 |
+
for i in range(len(values)):
|
489 |
+
for j in range(len(values)):
|
490 |
+
if mst[i, j] > 0:
|
491 |
+
points.append([i, j])
|
492 |
+
|
493 |
+
return points
|
494 |
+
|
495 |
+
################################################################################
|
496 |
+
|
497 |
+
def normalize_to_range(values, n):
|
498 |
+
|
499 |
+
min_val = min(values)
|
500 |
+
max_val = max(values)
|
501 |
+
|
502 |
+
range_val = max_val - min_val
|
503 |
+
|
504 |
+
normalized_values = [((value - min_val) / range_val * 2 * n) - n for value in values]
|
505 |
+
|
506 |
+
return normalized_values
|
507 |
+
|
508 |
+
################################################################################
|
509 |
+
|
510 |
+
def reduce_dimensionality_simple_pca(list_of_values, n_components=2):
|
511 |
+
|
512 |
+
'''
|
513 |
+
Reduces the dimensionality of the values using simple PCA
|
514 |
+
'''
|
515 |
+
|
516 |
+
reduced_values = []
|
517 |
+
|
518 |
+
for l in list_of_values:
|
519 |
+
|
520 |
+
norm_values = [round(v * len(l)) for v in normalize_to_range(l, (n_components+1) // 2)]
|
521 |
+
|
522 |
+
pca_values = Counter(norm_values).most_common()
|
523 |
+
pca_values = [vv[0] / len(l) for vv in pca_values]
|
524 |
+
pca_values = pca_values[:n_components]
|
525 |
+
pca_values = pca_values + [0] * (n_components - len(pca_values))
|
526 |
+
|
527 |
+
reduced_values.append(pca_values)
|
528 |
+
|
529 |
+
return reduced_values
|
530 |
+
|
531 |
+
################################################################################
|
532 |
+
|
533 |
+
def filter_and_replace_values(list_of_values,
|
534 |
+
threshold,
|
535 |
+
replace_value,
|
536 |
+
replace_above_threshold=False
|
537 |
+
):
|
538 |
+
|
539 |
+
array = np.array(list_of_values)
|
540 |
+
|
541 |
+
modified_array = np.copy(array)
|
542 |
+
|
543 |
+
if replace_above_threshold:
|
544 |
+
modified_array[modified_array > threshold] = replace_value
|
545 |
+
|
546 |
+
else:
|
547 |
+
modified_array[modified_array < threshold] = replace_value
|
548 |
+
|
549 |
+
return modified_array.tolist()
|
550 |
+
|
551 |
+
################################################################################
|
552 |
+
|
553 |
+
def find_shortest_constellation_path(points,
|
554 |
+
start_point_idx,
|
555 |
+
end_point_idx,
|
556 |
+
p=5,
|
557 |
+
return_path_length=False,
|
558 |
+
return_path_points=False,
|
559 |
+
):
|
560 |
+
|
561 |
+
"""
|
562 |
+
Finds the shortest path between two points of the points constellation
|
563 |
+
"""
|
564 |
+
|
565 |
+
points = np.array(points)
|
566 |
+
|
567 |
+
dist_matrix = distance_matrix(points, points, p=p)
|
568 |
+
|
569 |
+
mst = minimum_spanning_tree(dist_matrix).toarray()
|
570 |
+
|
571 |
+
G = nx.Graph()
|
572 |
+
|
573 |
+
for i in range(len(points)):
|
574 |
+
for j in range(len(points)):
|
575 |
+
if mst[i, j] > 0:
|
576 |
+
G.add_edge(i, j, weight=mst[i, j])
|
577 |
+
|
578 |
+
path = nx.shortest_path(G,
|
579 |
+
source=start_point_idx,
|
580 |
+
target=end_point_idx,
|
581 |
+
weight='weight'
|
582 |
+
)
|
583 |
+
|
584 |
+
path_length = nx.shortest_path_length(G,
|
585 |
+
source=start_point_idx,
|
586 |
+
target=end_point_idx,
|
587 |
+
weight='weight')
|
588 |
+
|
589 |
+
path_points = points[np.array(path)].tolist()
|
590 |
+
|
591 |
+
|
592 |
+
if return_path_points:
|
593 |
+
return path_points
|
594 |
+
|
595 |
+
if return_path_length:
|
596 |
+
return path_length
|
597 |
+
|
598 |
+
return path
|
599 |
+
|
600 |
+
################################################################################
|
601 |
+
# Core functions
|
602 |
+
################################################################################
|
603 |
+
|
604 |
+
def plot_ms_SONG(ms_song,
|
605 |
+
preview_length_in_notes=0,
|
606 |
+
block_lines_times_list = None,
|
607 |
+
plot_title='ms Song',
|
608 |
+
max_num_colors=129,
|
609 |
+
drums_color_num=128,
|
610 |
+
plot_size=(11,4),
|
611 |
+
note_height = 0.75,
|
612 |
+
show_grid_lines=False,
|
613 |
+
return_plt = False,
|
614 |
+
timings_multiplier=1,
|
615 |
+
save_plt='',
|
616 |
+
save_only_plt_image=True,
|
617 |
+
save_transparent=False
|
618 |
+
):
|
619 |
+
|
620 |
+
'''ms SONG plot'''
|
621 |
+
|
622 |
+
notes = [s for s in ms_song if s[0] == 'note']
|
623 |
+
|
624 |
+
if (len(max(notes, key=len)) != 7) and (len(min(notes, key=len)) != 7):
|
625 |
+
print('The song notes do not have patches information')
|
626 |
+
print('Ploease add patches to the notes in the song')
|
627 |
+
|
628 |
+
else:
|
629 |
+
|
630 |
+
start_times = [(s[1] * timings_multiplier) / 1000 for s in notes]
|
631 |
+
durations = [(s[2] * timings_multiplier) / 1000 for s in notes]
|
632 |
+
pitches = [s[4] for s in notes]
|
633 |
+
patches = [s[6] for s in notes]
|
634 |
+
|
635 |
+
colors = generate_colors(max_num_colors)
|
636 |
+
colors[drums_color_num] = (1, 1, 1)
|
637 |
+
|
638 |
+
pbl = (notes[preview_length_in_notes][1] * timings_multiplier) / 1000
|
639 |
+
|
640 |
+
fig, ax = plt.subplots(figsize=plot_size)
|
641 |
+
|
642 |
+
for start, duration, pitch, patch in zip(start_times, durations, pitches, patches):
|
643 |
+
rect = plt.Rectangle((start, pitch), duration, note_height, facecolor=colors[patch])
|
644 |
+
ax.add_patch(rect)
|
645 |
+
|
646 |
+
ax.set_xlim([min(start_times), max(add_arrays(start_times, durations))])
|
647 |
+
ax.set_ylim([min(pitches)-1, max(pitches)+1])
|
648 |
+
|
649 |
+
ax.set_facecolor('black')
|
650 |
+
fig.patch.set_facecolor('white')
|
651 |
+
|
652 |
+
if preview_length_in_notes > 0:
|
653 |
+
ax.axvline(x=pbl, c='white')
|
654 |
+
|
655 |
+
if block_lines_times_list:
|
656 |
+
for bl in block_lines_times_list:
|
657 |
+
ax.axvline(x=bl, c='white')
|
658 |
+
|
659 |
+
if show_grid_lines:
|
660 |
+
ax.grid(color='white')
|
661 |
+
|
662 |
+
plt.xlabel('Time (s)', c='black')
|
663 |
+
plt.ylabel('MIDI Pitch', c='black')
|
664 |
+
|
665 |
+
plt.title(plot_title)
|
666 |
+
|
667 |
+
if save_plt != '':
|
668 |
+
if save_only_plt_image:
|
669 |
+
plt.axis('off')
|
670 |
+
plt.title('')
|
671 |
+
plt.savefig(save_plt,
|
672 |
+
transparent=save_transparent,
|
673 |
+
bbox_inches='tight',
|
674 |
+
pad_inches=0,
|
675 |
+
facecolor='black'
|
676 |
+
)
|
677 |
+
plt.close()
|
678 |
+
|
679 |
+
else:
|
680 |
+
plt.savefig(save_plt)
|
681 |
+
plt.close()
|
682 |
+
|
683 |
+
if return_plt:
|
684 |
+
return fig
|
685 |
+
|
686 |
+
plt.show()
|
687 |
+
plt.close()
|
688 |
+
|
689 |
+
################################################################################
|
690 |
+
|
691 |
+
def plot_square_matrix_points(list_of_points,
|
692 |
+
list_of_points_colors,
|
693 |
+
plot_size=(7, 7),
|
694 |
+
point_size = 10,
|
695 |
+
show_grid_lines=False,
|
696 |
+
plot_title = 'Square Matrix Points Plot',
|
697 |
+
return_plt=False,
|
698 |
+
save_plt='',
|
699 |
+
save_only_plt_image=True,
|
700 |
+
save_transparent=False
|
701 |
+
):
|
702 |
+
|
703 |
+
'''Square matrix points plot'''
|
704 |
+
|
705 |
+
fig, ax = plt.subplots(figsize=plot_size)
|
706 |
+
|
707 |
+
ax.set_facecolor('black')
|
708 |
+
|
709 |
+
if show_grid_lines:
|
710 |
+
ax.grid(color='white')
|
711 |
+
|
712 |
+
plt.xlabel('Time Step', c='black')
|
713 |
+
plt.ylabel('MIDI Pitch', c='black')
|
714 |
+
|
715 |
+
plt.title(plot_title)
|
716 |
+
|
717 |
+
plt.scatter([p[0] for p in list_of_points],
|
718 |
+
[p[1] for p in list_of_points],
|
719 |
+
c=list_of_points_colors,
|
720 |
+
s=point_size
|
721 |
+
)
|
722 |
+
|
723 |
+
if save_plt != '':
|
724 |
+
if save_only_plt_image:
|
725 |
+
plt.axis('off')
|
726 |
+
plt.title('')
|
727 |
+
plt.savefig(save_plt,
|
728 |
+
transparent=save_transparent,
|
729 |
+
bbox_inches='tight',
|
730 |
+
pad_inches=0,
|
731 |
+
facecolor='black'
|
732 |
+
)
|
733 |
+
plt.close()
|
734 |
+
|
735 |
+
else:
|
736 |
+
plt.savefig(save_plt)
|
737 |
+
plt.close()
|
738 |
+
|
739 |
+
if return_plt:
|
740 |
+
return fig
|
741 |
+
|
742 |
+
plt.show()
|
743 |
+
plt.close()
|
744 |
+
|
745 |
+
################################################################################
|
746 |
+
|
747 |
+
def plot_cosine_similarities(lists_of_values,
|
748 |
+
plot_size=(7, 7),
|
749 |
+
save_plot=''
|
750 |
+
):
|
751 |
+
|
752 |
+
"""
|
753 |
+
Cosine similarities plot
|
754 |
+
"""
|
755 |
+
|
756 |
+
cos_sim = metrics.pairwise_distances(lists_of_values, metric='cosine')
|
757 |
+
|
758 |
+
plt.figure(figsize=plot_size)
|
759 |
+
|
760 |
+
plt.imshow(cos_sim, cmap="inferno", interpolation="nearest")
|
761 |
+
|
762 |
+
im_ratio = cos_sim.shape[0] / cos_sim.shape[1]
|
763 |
+
|
764 |
+
plt.colorbar(fraction=0.046 * im_ratio, pad=0.04)
|
765 |
+
|
766 |
+
plt.xlabel("Index")
|
767 |
+
plt.ylabel("Index")
|
768 |
+
|
769 |
+
plt.tight_layout()
|
770 |
+
|
771 |
+
if save_plot != '':
|
772 |
+
plt.savefig(save_plot, bbox_inches="tight")
|
773 |
+
plt.close()
|
774 |
+
|
775 |
+
plt.show()
|
776 |
+
plt.close()
|
777 |
+
|
778 |
+
################################################################################
|
779 |
+
|
780 |
+
def plot_points_with_mst_lines(points,
|
781 |
+
points_labels,
|
782 |
+
points_mst_edges,
|
783 |
+
plot_size=(20, 20),
|
784 |
+
labels_size=24,
|
785 |
+
save_plot=''
|
786 |
+
):
|
787 |
+
|
788 |
+
"""
|
789 |
+
Plots 2D points with labels and MST lines.
|
790 |
+
"""
|
791 |
+
|
792 |
+
plt.figure(figsize=plot_size)
|
793 |
+
|
794 |
+
for i, label in enumerate(points_labels):
|
795 |
+
plt.scatter(points[i][0], points[i][1])
|
796 |
+
plt.annotate(label, (points[i][0], points[i][1]), fontsize=labels_size)
|
797 |
+
|
798 |
+
for edge in points_mst_edges:
|
799 |
+
i, j = edge
|
800 |
+
plt.plot([points[i][0], points[j][0]], [points[i][1], points[j][1]], 'k-', alpha=0.5)
|
801 |
+
|
802 |
+
plt.title('Points Map with MST Lines', fontsize=labels_size)
|
803 |
+
plt.xlabel('X-axis', fontsize=labels_size)
|
804 |
+
plt.ylabel('Y-axis', fontsize=labels_size)
|
805 |
+
|
806 |
+
if save_plot != '':
|
807 |
+
plt.savefig(save_plot, bbox_inches="tight")
|
808 |
+
plt.close()
|
809 |
+
|
810 |
+
plt.show()
|
811 |
+
|
812 |
+
plt.close()
|
813 |
+
|
814 |
+
################################################################################
|
815 |
+
|
816 |
+
def plot_points_constellation(points,
|
817 |
+
points_labels,
|
818 |
+
p=5,
|
819 |
+
plot_size=(15, 15),
|
820 |
+
labels_size=12,
|
821 |
+
show_grid=False,
|
822 |
+
save_plot=''
|
823 |
+
):
|
824 |
+
|
825 |
+
"""
|
826 |
+
Plots 2D points constellation
|
827 |
+
"""
|
828 |
+
|
829 |
+
points = np.array(points)
|
830 |
+
|
831 |
+
dist_matrix = distance_matrix(points, points, p=p)
|
832 |
+
|
833 |
+
mst = minimum_spanning_tree(dist_matrix).toarray()
|
834 |
+
|
835 |
+
plt.figure(figsize=plot_size)
|
836 |
+
|
837 |
+
plt.scatter(points[:, 0], points[:, 1], color='blue')
|
838 |
+
|
839 |
+
for i, label in enumerate(points_labels):
|
840 |
+
plt.annotate(label, (points[i, 0], points[i, 1]),
|
841 |
+
textcoords="offset points",
|
842 |
+
xytext=(0, 10),
|
843 |
+
ha='center',
|
844 |
+
fontsize=labels_size
|
845 |
+
)
|
846 |
+
|
847 |
+
for i in range(len(points)):
|
848 |
+
for j in range(len(points)):
|
849 |
+
if mst[i, j] > 0:
|
850 |
+
plt.plot([points[i, 0], points[j, 0]], [points[i, 1], points[j, 1]], 'k--')
|
851 |
+
|
852 |
+
plt.xlabel('X-axis', fontsize=labels_size)
|
853 |
+
plt.ylabel('Y-axis', fontsize=labels_size)
|
854 |
+
plt.title('2D Coordinates with Minimum Spanning Tree', fontsize=labels_size)
|
855 |
+
|
856 |
+
plt.grid(show_grid)
|
857 |
+
|
858 |
+
if save_plot != '':
|
859 |
+
plt.savefig(save_plot, bbox_inches="tight")
|
860 |
+
plt.close()
|
861 |
+
|
862 |
+
plt.show()
|
863 |
+
|
864 |
+
plt.close()
|
865 |
+
|
866 |
+
################################################################################
|
867 |
+
|
868 |
+
def binary_matrix_to_images(matrix,
|
869 |
+
step,
|
870 |
+
overlap,
|
871 |
+
output_folder='./Dataset/',
|
872 |
+
output_img_prefix='image',
|
873 |
+
output_img_ext='.png',
|
874 |
+
save_to_array=False,
|
875 |
+
verbose=True
|
876 |
+
):
|
877 |
+
|
878 |
+
if not save_to_array:
|
879 |
+
|
880 |
+
if verbose:
|
881 |
+
print('=' * 70)
|
882 |
+
print('Checking output folder dir...')
|
883 |
+
|
884 |
+
os.makedirs(os.path.dirname(output_folder), exist_ok=True)
|
885 |
+
|
886 |
+
if verbose:
|
887 |
+
print('Done!')
|
888 |
+
|
889 |
+
if verbose:
|
890 |
+
print('=' * 70)
|
891 |
+
print('Writing images...')
|
892 |
+
|
893 |
+
matrix = np.array(matrix, dtype=np.uint8)
|
894 |
+
|
895 |
+
image_array = []
|
896 |
+
|
897 |
+
for i in range(0, max(1, matrix.shape[0]), overlap):
|
898 |
+
|
899 |
+
submatrix = matrix[i:i+step, :]
|
900 |
+
|
901 |
+
if submatrix.shape[0] < 128:
|
902 |
+
zeros_array = np.zeros((128-submatrix.shape[0], 128))
|
903 |
+
submatrix = np.vstack((submatrix, zeros_array))
|
904 |
+
|
905 |
+
img = Image.fromarray(submatrix * 255).convert('1')
|
906 |
+
|
907 |
+
if save_to_array:
|
908 |
+
image_array.append(np.array(img))
|
909 |
+
|
910 |
+
else:
|
911 |
+
img.save(output_folder + output_img_prefix + '_' + str(matrix.shape[1]) + '_' + str(i).zfill(7) + output_img_ext)
|
912 |
+
|
913 |
+
if verbose:
|
914 |
+
print('Done!')
|
915 |
+
print('=' * 70)
|
916 |
+
print('Saved', (matrix.shape[0] // min(step, overlap))+1, 'imges!')
|
917 |
+
print('=' * 70)
|
918 |
+
|
919 |
+
if save_to_array:
|
920 |
+
return np.array(image_array).tolist()
|
921 |
+
|
922 |
+
################################################################################
|
923 |
+
|
924 |
+
def images_to_binary_matrix(list_of_images):
|
925 |
+
|
926 |
+
image_array = np.array(list_of_images)
|
927 |
+
|
928 |
+
original_matrix = []
|
929 |
+
|
930 |
+
for img in image_array:
|
931 |
+
|
932 |
+
submatrix = np.array(img)
|
933 |
+
original_matrix.extend(submatrix.tolist())
|
934 |
+
|
935 |
+
return original_matrix
|
936 |
+
|
937 |
+
################################################################################
|
938 |
+
|
939 |
+
def square_image_matrix(image_matrix,
|
940 |
+
matrix_size=128,
|
941 |
+
num_pca_components=5,
|
942 |
+
filter_out_zero_rows=False,
|
943 |
+
return_square_matrix_points=False
|
944 |
+
):
|
945 |
+
|
946 |
+
"""
|
947 |
+
Reduces an arbitrary image matrix to a square image matrix
|
948 |
+
"""
|
949 |
+
|
950 |
+
matrix = np.array(image_matrix)
|
951 |
+
|
952 |
+
if filter_out_zero_rows:
|
953 |
+
matrix = matrix[~np.all(matrix == 0, axis=1)]
|
954 |
+
|
955 |
+
target_rows = matrix_size
|
956 |
+
|
957 |
+
rows_per_group = matrix.shape[0] // target_rows
|
958 |
+
|
959 |
+
compressed_matrix = np.zeros((target_rows, matrix.shape[1]), dtype=np.int32)
|
960 |
+
|
961 |
+
for i in range(target_rows):
|
962 |
+
start_row = i * rows_per_group
|
963 |
+
end_row = (i + 1) * rows_per_group
|
964 |
+
group = matrix[start_row:end_row, :]
|
965 |
+
|
966 |
+
pca = PCA(n_components=num_pca_components)
|
967 |
+
pca.fit(group)
|
968 |
+
|
969 |
+
principal_component = np.mean(pca.components_, axis=0)
|
970 |
+
contributions = np.dot(group, principal_component)
|
971 |
+
selected_row_index = np.argmax(contributions)
|
972 |
+
|
973 |
+
compressed_matrix[i, :] = group[selected_row_index, :]
|
974 |
+
|
975 |
+
if return_square_matrix_points:
|
976 |
+
filtered_matrix = compressed_matrix[~np.all(compressed_matrix == 0, axis=1)]
|
977 |
+
|
978 |
+
row_indexes, col_indexes = np.where(filtered_matrix != 0)
|
979 |
+
points = np.column_stack((row_indexes, filtered_matrix[row_indexes, col_indexes])).tolist()
|
980 |
+
|
981 |
+
return points
|
982 |
+
|
983 |
+
else:
|
984 |
+
return compressed_matrix.tolist()
|
985 |
+
|
986 |
+
################################################################################
|
987 |
+
|
988 |
+
def image_matrix_to_images(image_matrix,
|
989 |
+
step,
|
990 |
+
overlap,
|
991 |
+
num_img_channels=3,
|
992 |
+
output_folder='./Dataset/',
|
993 |
+
output_img_prefix='image',
|
994 |
+
output_img_ext='.png',
|
995 |
+
save_to_array=False,
|
996 |
+
verbose=True
|
997 |
+
):
|
998 |
+
|
999 |
+
if num_img_channels > 1:
|
1000 |
+
n_mat_channels = 3
|
1001 |
+
|
1002 |
+
else:
|
1003 |
+
n_mat_channels = 1
|
1004 |
+
|
1005 |
+
if not save_to_array:
|
1006 |
+
|
1007 |
+
if verbose:
|
1008 |
+
print('=' * 70)
|
1009 |
+
print('Checking output folder dir...')
|
1010 |
+
|
1011 |
+
os.makedirs(os.path.dirname(output_folder), exist_ok=True)
|
1012 |
+
|
1013 |
+
if verbose:
|
1014 |
+
print('Done!')
|
1015 |
+
|
1016 |
+
if verbose:
|
1017 |
+
print('=' * 70)
|
1018 |
+
print('Writing images...')
|
1019 |
+
|
1020 |
+
matrix = np.array(image_matrix)
|
1021 |
+
|
1022 |
+
image_array = []
|
1023 |
+
|
1024 |
+
for i in range(0, max(1, matrix.shape[0]), overlap):
|
1025 |
+
|
1026 |
+
submatrix = matrix[i:i+step, :]
|
1027 |
+
|
1028 |
+
if submatrix.shape[0] < 128:
|
1029 |
+
zeros_array = np.zeros((128-submatrix.shape[0], 128))
|
1030 |
+
submatrix = np.vstack((submatrix, zeros_array))
|
1031 |
+
|
1032 |
+
if n_mat_channels == 3:
|
1033 |
+
|
1034 |
+
r = (submatrix // (256*256)) % 256
|
1035 |
+
g = (submatrix // 256) % 256
|
1036 |
+
b = submatrix % 256
|
1037 |
+
|
1038 |
+
rgb_image = np.stack((r, g, b), axis=-1).astype(np.uint8)
|
1039 |
+
img = Image.fromarray(rgb_image, 'RGB')
|
1040 |
+
|
1041 |
+
else:
|
1042 |
+
grayscale_image = submatrix.astype(np.uint8)
|
1043 |
+
img = Image.fromarray(grayscale_image, 'L')
|
1044 |
+
|
1045 |
+
if save_to_array:
|
1046 |
+
image_array.append(np.array(img))
|
1047 |
+
|
1048 |
+
else:
|
1049 |
+
img.save(output_folder + output_img_prefix + '_' + str(matrix.shape[1]) + '_' + str(i).zfill(7) + output_img_ext)
|
1050 |
+
|
1051 |
+
if verbose:
|
1052 |
+
print('Done!')
|
1053 |
+
print('=' * 70)
|
1054 |
+
print('Saved', (matrix.shape[0] // min(step, overlap))+1, 'imges!')
|
1055 |
+
print('=' * 70)
|
1056 |
+
|
1057 |
+
if save_to_array:
|
1058 |
+
return np.array(image_array).tolist()
|
1059 |
+
|
1060 |
+
################################################################################
|
1061 |
+
|
1062 |
+
def images_to_image_matrix(list_of_images,
|
1063 |
+
num_img_channels=3
|
1064 |
+
):
|
1065 |
+
|
1066 |
+
if num_img_channels > 1:
|
1067 |
+
n_mat_channels = 3
|
1068 |
+
|
1069 |
+
else:
|
1070 |
+
n_mat_channels = 1
|
1071 |
+
|
1072 |
+
image_array = np.array(list_of_images)
|
1073 |
+
|
1074 |
+
original_matrix = []
|
1075 |
+
|
1076 |
+
for img in image_array:
|
1077 |
+
|
1078 |
+
if num_img_channels == 3:
|
1079 |
+
|
1080 |
+
rgb_array = np.array(img)
|
1081 |
+
|
1082 |
+
matrix = (rgb_array[..., 0].astype(np.int64) * 256*256 +
|
1083 |
+
rgb_array[..., 1].astype(np.int64) * 256 +
|
1084 |
+
rgb_array[..., 2].astype(np.int64))
|
1085 |
+
|
1086 |
+
else:
|
1087 |
+
matrix = np.array(img)
|
1088 |
+
|
1089 |
+
original_matrix.extend(matrix)
|
1090 |
+
|
1091 |
+
return original_matrix
|
1092 |
+
|
1093 |
+
################################################################################
|
1094 |
+
# [WIP] Future dev functions
|
1095 |
+
################################################################################
|
1096 |
+
|
1097 |
+
'''
|
1098 |
+
import umap
|
1099 |
+
|
1100 |
+
def reduce_dimensionality_umap(list_of_values,
|
1101 |
+
n_comp=2,
|
1102 |
+
n_neighbors=15,
|
1103 |
+
):
|
1104 |
+
|
1105 |
+
"""
|
1106 |
+
Reduces the dimensionality of the values using UMAP.
|
1107 |
+
"""
|
1108 |
+
|
1109 |
+
vals = np.array(list_of_values)
|
1110 |
+
|
1111 |
+
umap_reducer = umap.UMAP(n_components=n_comp,
|
1112 |
+
n_neighbors=n_neighbors,
|
1113 |
+
n_epochs=5000,
|
1114 |
+
verbose=True
|
1115 |
+
)
|
1116 |
+
|
1117 |
+
reduced_vals = umap_reducer.fit_transform(vals)
|
1118 |
+
|
1119 |
+
return reduced_vals.tolist()
|
1120 |
+
'''
|
1121 |
+
|
1122 |
+
################################################################################
|
1123 |
+
|
1124 |
+
'''
|
1125 |
+
import alphashape
|
1126 |
+
from shapely.geometry import Point
|
1127 |
+
from matplotlib.tri import Triangulation, LinearTriInterpolator
|
1128 |
+
from scipy.stats import zscore
|
1129 |
+
|
1130 |
+
#===============================================================================
|
1131 |
+
|
1132 |
+
coordinates = points
|
1133 |
+
|
1134 |
+
dist_matrix = minkowski_distance_matrix(coordinates, p=3) # You can change the value of p as needed
|
1135 |
+
|
1136 |
+
# Centering matrix
|
1137 |
+
n = dist_matrix.shape[0]
|
1138 |
+
H = np.eye(n) - np.ones((n, n)) / n
|
1139 |
+
|
1140 |
+
# Apply double centering
|
1141 |
+
B = -0.5 * H @ dist_matrix**2 @ H
|
1142 |
+
|
1143 |
+
# Eigen decomposition
|
1144 |
+
eigvals, eigvecs = np.linalg.eigh(B)
|
1145 |
+
|
1146 |
+
# Sort eigenvalues and eigenvectors
|
1147 |
+
idx = np.argsort(eigvals)[::-1]
|
1148 |
+
eigvals = eigvals[idx]
|
1149 |
+
eigvecs = eigvecs[:, idx]
|
1150 |
+
|
1151 |
+
# Select the top 2 eigenvectors
|
1152 |
+
X_transformed = eigvecs[:, :2] * np.sqrt(eigvals[:2])
|
1153 |
+
|
1154 |
+
#===============================================================================
|
1155 |
+
|
1156 |
+
src_points = X_transformed
|
1157 |
+
src_values = np.array([[p[1]] for p in points]) #np.random.rand(X_transformed.shape[0])
|
1158 |
+
|
1159 |
+
#===============================================================================
|
1160 |
+
|
1161 |
+
# Normalize the points to the range [0, 1]
|
1162 |
+
scaler = MinMaxScaler()
|
1163 |
+
points_normalized = scaler.fit_transform(src_points)
|
1164 |
+
|
1165 |
+
values_normalized = custom_normalize(src_values)
|
1166 |
+
|
1167 |
+
# Remove outliers based on z-score
|
1168 |
+
z_scores = np.abs(zscore(points_normalized, axis=0))
|
1169 |
+
filtered_points = points_normalized[(z_scores < 3).all(axis=1)]
|
1170 |
+
filtered_values = values_normalized[(z_scores < 3).all(axis=1)]
|
1171 |
+
|
1172 |
+
# Compute the concave hull (alpha shape)
|
1173 |
+
alpha = 8 # Adjust alpha as needed
|
1174 |
+
hull = alphashape.alphashape(filtered_points, alpha)
|
1175 |
+
|
1176 |
+
# Create a triangulation
|
1177 |
+
tri = Triangulation(filtered_points[:, 0], filtered_points[:, 1])
|
1178 |
+
|
1179 |
+
# Interpolate the values on the triangulation
|
1180 |
+
interpolator = LinearTriInterpolator(tri, filtered_values[:, 0])
|
1181 |
+
xi, yi = np.meshgrid(np.linspace(0, 1, 100), np.linspace(0, 1, 100))
|
1182 |
+
zi = interpolator(xi, yi)
|
1183 |
+
|
1184 |
+
# Mask out points outside the concave hull
|
1185 |
+
mask = np.array([hull.contains(Point(x, y)) for x, y in zip(xi.flatten(), yi.flatten())])
|
1186 |
+
zi = np.ma.array(zi, mask=~mask.reshape(zi.shape))
|
1187 |
+
|
1188 |
+
# Plot the filled contour based on the interpolated values
|
1189 |
+
plt.contourf(xi, yi, zi, levels=50, cmap='viridis')
|
1190 |
+
|
1191 |
+
# Plot the original points
|
1192 |
+
#plt.scatter(filtered_points[:, 0], filtered_points[:, 1], c=filtered_values, edgecolors='k')
|
1193 |
+
|
1194 |
+
plt.title('Filled Contour Plot with Original Values')
|
1195 |
+
plt.xlabel('X-axis')
|
1196 |
+
plt.ylabel('Y-axis')
|
1197 |
+
plt.colorbar(label='Value')
|
1198 |
+
plt.show()
|
1199 |
+
'''
|
1200 |
+
|
1201 |
+
################################################################################
|
1202 |
+
#
|
1203 |
+
# This is the end of TPLOTS Python modules
|
1204 |
+
#
|
1205 |
+
################################################################################
|
midi_images_solo_piano_dataset_maker.py
ADDED
@@ -0,0 +1,330 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# -*- coding: utf-8 -*-
|
2 |
+
"""MIDI_Images_Solo_Piano_Dataset_Maker.ipynb
|
3 |
+
|
4 |
+
Automatically generated by Colab.
|
5 |
+
|
6 |
+
Original file is located at
|
7 |
+
https://colab.research.google.com/drive/15E6o3Y1xPific5RtIZ-1CneHQhts3eEr
|
8 |
+
|
9 |
+
# MIDI Images Solo Piano Dataset Maker (ver. 1.0)
|
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 |
+
|
26 |
+
# @title Install dependecies
|
27 |
+
!git clone --depth 1 https://github.com/asigalov61/tegridy-tools
|
28 |
+
|
29 |
+
# Commented out IPython magic to ensure Python compatibility.
|
30 |
+
#@title Import all needed modules
|
31 |
+
|
32 |
+
print('=' * 70)
|
33 |
+
print('Loading core modules...')
|
34 |
+
print('Please wait...')
|
35 |
+
print('=' * 70)
|
36 |
+
|
37 |
+
import os
|
38 |
+
import copy
|
39 |
+
import math
|
40 |
+
import statistics
|
41 |
+
import random
|
42 |
+
import pickle
|
43 |
+
import shutil
|
44 |
+
from itertools import groupby
|
45 |
+
from collections import Counter
|
46 |
+
from sklearn.metrics import pairwise_distances
|
47 |
+
from sklearn import metrics
|
48 |
+
from joblib import Parallel, delayed, parallel_config
|
49 |
+
import numpy as np
|
50 |
+
from tqdm import tqdm
|
51 |
+
from PIL import Image
|
52 |
+
import matplotlib.pyplot as plt
|
53 |
+
|
54 |
+
print('Done!')
|
55 |
+
print('=' * 70)
|
56 |
+
print('Creating I/O dirs...')
|
57 |
+
|
58 |
+
if not os.path.exists('/content/Dataset'):
|
59 |
+
os.makedirs('/content/Dataset')
|
60 |
+
|
61 |
+
print('Done!')
|
62 |
+
print('=' * 70)
|
63 |
+
print('Loading tegridy-tools modules...')
|
64 |
+
print('=' * 70)
|
65 |
+
|
66 |
+
# %cd /content/tegridy-tools/tegridy-tools
|
67 |
+
|
68 |
+
import TMIDIX
|
69 |
+
import TMELODIES
|
70 |
+
import TPLOTS
|
71 |
+
import HaystackSearch
|
72 |
+
|
73 |
+
# %cd /content/
|
74 |
+
|
75 |
+
print('=' * 70)
|
76 |
+
print('Done!')
|
77 |
+
print('=' * 70)
|
78 |
+
|
79 |
+
"""# (DOWNLOAD SAMPLE MIDI DATASET)"""
|
80 |
+
|
81 |
+
# Commented out IPython magic to ensure Python compatibility.
|
82 |
+
# @title Download sample MIDI dataset (POP909)
|
83 |
+
# %cd /content/Dataset/
|
84 |
+
!git clone --depth 1 https://github.com/music-x-lab/POP909-Dataset
|
85 |
+
# %cd /content/
|
86 |
+
|
87 |
+
#@title Save file list
|
88 |
+
###########
|
89 |
+
|
90 |
+
print('=' * 70)
|
91 |
+
print('Loading MIDI files...')
|
92 |
+
print('This may take a while on a large dataset in particular...')
|
93 |
+
|
94 |
+
dataset_addr = '/content/Dataset/'
|
95 |
+
|
96 |
+
# os.chdir(dataset_addr)
|
97 |
+
filez = list()
|
98 |
+
for (dirpath, dirnames, filenames) in os.walk(dataset_addr):
|
99 |
+
filez += [os.path.join(dirpath, file) for file in filenames if file.endswith('.mid') or file.endswith('.midi') or file.endswith('.kar')]
|
100 |
+
print('=' * 70)
|
101 |
+
|
102 |
+
if filez == []:
|
103 |
+
print('Could not find any MIDI files. Please check Dataset dir...')
|
104 |
+
print('=' * 70)
|
105 |
+
|
106 |
+
print('Randomizing file list...')
|
107 |
+
random.shuffle(filez)
|
108 |
+
print('Done!')
|
109 |
+
print('=' * 70)
|
110 |
+
print('Total found MIDI files:', len(filez))
|
111 |
+
print('=' * 70)
|
112 |
+
|
113 |
+
TMIDIX.Tegridy_Any_Pickle_File_Writer(filez, 'filez')
|
114 |
+
|
115 |
+
print('=' * 70)
|
116 |
+
|
117 |
+
"""# (LOAD TMIDIX MIDI PROCESSOR)"""
|
118 |
+
|
119 |
+
#@title Load TMIDIX MIDI processor
|
120 |
+
|
121 |
+
print('=' * 70)
|
122 |
+
print('TMIDIX MIDI Processor')
|
123 |
+
print('=' * 70)
|
124 |
+
print('Loading...')
|
125 |
+
|
126 |
+
###########
|
127 |
+
|
128 |
+
def TMIDIX_MIDI_Processor(midi_file):
|
129 |
+
|
130 |
+
fn = os.path.basename(midi_file)
|
131 |
+
fn1 = fn.split('.mid')[0]
|
132 |
+
|
133 |
+
try:
|
134 |
+
|
135 |
+
#=======================================================
|
136 |
+
# START PROCESSING
|
137 |
+
|
138 |
+
raw_score = TMIDIX.midi2single_track_ms_score(midi_file)
|
139 |
+
|
140 |
+
escore_notes = TMIDIX.advanced_score_processor(raw_score, return_enhanced_score_notes=True)[0]
|
141 |
+
|
142 |
+
escore_notes = TMIDIX.augment_enhanced_score_notes(escore_notes, timings_divider=256)
|
143 |
+
|
144 |
+
sp_escore_notes = TMIDIX.recalculate_score_timings(TMIDIX.solo_piano_escore_notes(escore_notes, keep_drums=False))
|
145 |
+
|
146 |
+
if sp_escore_notes:
|
147 |
+
|
148 |
+
bmatrix = TMIDIX.escore_notes_to_binary_matrix(sp_escore_notes)
|
149 |
+
|
150 |
+
return [fn1, bmatrix]
|
151 |
+
|
152 |
+
else:
|
153 |
+
return [fn1, []]
|
154 |
+
|
155 |
+
#=======================================================
|
156 |
+
|
157 |
+
except Exception as ex:
|
158 |
+
print('WARNING !!!')
|
159 |
+
print('=' * 70)
|
160 |
+
print('Bad MIDI:', midi_file)
|
161 |
+
print('Error detected:', ex)
|
162 |
+
print('=' * 70)
|
163 |
+
return None
|
164 |
+
|
165 |
+
print('Done!')
|
166 |
+
print('=' * 70)
|
167 |
+
|
168 |
+
"""# (PROCESS MIDIs)"""
|
169 |
+
|
170 |
+
#@title Process MIDIs with TMIDIX MIDI processor
|
171 |
+
output_folder = "/content/MIDI-Images/" # @param {"type":"string"}
|
172 |
+
|
173 |
+
NUMBER_OF_PARALLEL_JOBS = 4 # Number of parallel jobs
|
174 |
+
NUMBER_OF_FILES_PER_ITERATION = 4 # Number of files to queue for each parallel iteration
|
175 |
+
SAVE_EVERY_NUMBER_OF_ITERATIONS = 128 # Save every 2560 files
|
176 |
+
|
177 |
+
print('=' * 70)
|
178 |
+
print('TMIDIX MIDI Processor')
|
179 |
+
print('=' * 70)
|
180 |
+
print('Starting up...')
|
181 |
+
print('=' * 70)
|
182 |
+
|
183 |
+
###########
|
184 |
+
|
185 |
+
melody_chords_f = []
|
186 |
+
|
187 |
+
files_count = 0
|
188 |
+
|
189 |
+
print('Processing MIDI files...')
|
190 |
+
print('Please wait...')
|
191 |
+
print('=' * 70)
|
192 |
+
|
193 |
+
for i in tqdm(range(0, len(filez), NUMBER_OF_FILES_PER_ITERATION)):
|
194 |
+
|
195 |
+
with parallel_config(backend='threading', n_jobs=NUMBER_OF_PARALLEL_JOBS, verbose = 0):
|
196 |
+
|
197 |
+
output = Parallel(n_jobs=NUMBER_OF_PARALLEL_JOBS, verbose=0)(delayed(TMIDIX_MIDI_Processor)(f) for f in filez[i:i+NUMBER_OF_FILES_PER_ITERATION])
|
198 |
+
|
199 |
+
for o in output:
|
200 |
+
|
201 |
+
if o is not None:
|
202 |
+
melody_chords_f.append(o)
|
203 |
+
|
204 |
+
if i % (NUMBER_OF_FILES_PER_ITERATION * SAVE_EVERY_NUMBER_OF_ITERATIONS) == 0 and i != 0:
|
205 |
+
|
206 |
+
print('SAVING !!!')
|
207 |
+
print('=' * 70)
|
208 |
+
print('Saving processed files...')
|
209 |
+
files_count += len(melody_chords_f)
|
210 |
+
print('=' * 70)
|
211 |
+
print('Processed so far:', files_count, 'out of', len(filez), '===', files_count / len(filez), 'good files ratio')
|
212 |
+
print('=' * 70)
|
213 |
+
print('Writing images...')
|
214 |
+
print('Please wait...')
|
215 |
+
|
216 |
+
for mat in melody_chords_f:
|
217 |
+
|
218 |
+
if mat[1]:
|
219 |
+
|
220 |
+
TPLOTS.binary_matrix_to_images(mat[1],
|
221 |
+
128,
|
222 |
+
32,
|
223 |
+
output_folder=output_folder+str(mat[0])+'/',
|
224 |
+
output_img_prefix=str(mat[0]),
|
225 |
+
output_img_ext='.png',
|
226 |
+
verbose=False
|
227 |
+
)
|
228 |
+
|
229 |
+
print('Done!')
|
230 |
+
print('=' * 70)
|
231 |
+
melody_chords_f = []
|
232 |
+
|
233 |
+
print('SAVING !!!')
|
234 |
+
print('=' * 70)
|
235 |
+
print('Saving processed files...')
|
236 |
+
files_count += len(melody_chords_f)
|
237 |
+
print('=' * 70)
|
238 |
+
print('Processed so far:', files_count, 'out of', len(filez), '===', files_count / len(filez), 'good files ratio')
|
239 |
+
print('=' * 70)
|
240 |
+
print('Writing images...')
|
241 |
+
print('Please wait...')
|
242 |
+
|
243 |
+
for mat in melody_chords_f:
|
244 |
+
|
245 |
+
if mat[1]:
|
246 |
+
|
247 |
+
TPLOTS.binary_matrix_to_images(mat[1],
|
248 |
+
128,
|
249 |
+
32,
|
250 |
+
output_folder=output_folder+str(mat[0])+'/',
|
251 |
+
output_img_prefix=str(mat[0]),
|
252 |
+
output_img_ext='.png',
|
253 |
+
verbose=False
|
254 |
+
)
|
255 |
+
|
256 |
+
print('Done!')
|
257 |
+
print('=' * 70)
|
258 |
+
|
259 |
+
"""# (LOAD IMAGES)"""
|
260 |
+
|
261 |
+
#@title Load created MIDI images
|
262 |
+
full_path_to_metadata_pickle_files = "/content/MIDI-Images" #@param {type:"string"}
|
263 |
+
|
264 |
+
print('=' * 70)
|
265 |
+
print('MIDI Images Reader')
|
266 |
+
print('=' * 70)
|
267 |
+
print('Searching for images...')
|
268 |
+
|
269 |
+
filez = list()
|
270 |
+
for (dirpath, dirnames, filenames) in os.walk(full_path_to_metadata_pickle_files):
|
271 |
+
filez += [os.path.join(dirpath, file) for file in filenames if file.endswith('.png')]
|
272 |
+
print('=' * 70)
|
273 |
+
|
274 |
+
filez.sort()
|
275 |
+
|
276 |
+
print('Found', len(filez), 'images!')
|
277 |
+
print('=' * 70)
|
278 |
+
print('Reading images...')
|
279 |
+
print('Please wait...')
|
280 |
+
print('=' * 70)
|
281 |
+
|
282 |
+
fidx = 0
|
283 |
+
|
284 |
+
all_read_images = []
|
285 |
+
|
286 |
+
for img in tqdm(filez):
|
287 |
+
|
288 |
+
img = Image.open(img)
|
289 |
+
|
290 |
+
img_arr = np.array(img).tolist()
|
291 |
+
|
292 |
+
all_read_images.append(img_arr)
|
293 |
+
|
294 |
+
fidx += 1
|
295 |
+
|
296 |
+
print('Done!')
|
297 |
+
print('=' * 70)
|
298 |
+
print('Loaded', fidx, 'images!')
|
299 |
+
print('=' * 70)
|
300 |
+
print('Done!')
|
301 |
+
print('=' * 70)
|
302 |
+
|
303 |
+
"""# (TEST IMAGES)"""
|
304 |
+
|
305 |
+
# @title Test created MIDI images
|
306 |
+
|
307 |
+
print('=' * 70)
|
308 |
+
|
309 |
+
image = random.choice(all_read_images)
|
310 |
+
|
311 |
+
escore = TMIDIX.binary_matrix_to_original_escore_notes(image)
|
312 |
+
|
313 |
+
output_score, patches, overflow_patches = TMIDIX.patch_enhanced_score_notes(escore)
|
314 |
+
|
315 |
+
detailed_stats = TMIDIX.Tegridy_ms_SONG_to_MIDI_Converter(output_score,
|
316 |
+
output_signature = 'MIDI Images',
|
317 |
+
output_file_name = '/content/MIDI-Images-Composition',
|
318 |
+
track_name='Project Los Angeles',
|
319 |
+
list_of_MIDI_patches=patches,
|
320 |
+
timings_multiplier=256
|
321 |
+
)
|
322 |
+
|
323 |
+
print('=' * 70)
|
324 |
+
|
325 |
+
"""# (ZIP IMAGES)"""
|
326 |
+
|
327 |
+
# @title Zip created MIDI images
|
328 |
+
!zip -9 -r POP909_MIDI_Images_128_128_32_BW.zip MIDI-Images/ > /dev/null
|
329 |
+
|
330 |
+
"""# Congrats! You did it! :)"""
|