#N canvas 321 38 941 662 12;
#X obj 464 555 spigot;
#X obj 385 555 spigot;
#N canvas 906 141 305 506 synth 0;
#X obj 72 159 * 12;
#X obj 72 184 div 7;
#X obj 72 234 mtof;
#X obj 72 339 osc~;
#X obj 72 364 cos~;
#X obj 141 339 line~;
#X obj 141 364 *~;
#X obj 141 389 lop~ 500;
#X obj 72 389 *~;
#X obj 72 461 dac~;
#X obj 174 261 dbtorms;
#X obj 141 287 f;
#X obj 174 97 inlet;
#X obj 72 412 hip~ 5;
#X msg 141 314 \$1 \, 0 200;
#X obj 141 263 b;
#X obj 72 97 inlet;
#X obj 72 129 t f;
#X obj 72 209 + 48;
#X connect 0 0 1 0;
#X connect 1 0 18 0;
#X connect 2 0 15 0;
#X connect 2 0 3 0;
#X connect 3 0 4 0;
#X connect 4 0 8 0;
#X connect 5 0 6 0;
#X connect 5 0 6 1;
#X connect 6 0 7 0;
#X connect 7 0 8 1;
#X connect 8 0 13 0;
#X connect 10 0 11 1;
#X connect 11 0 14 0;
#X connect 12 0 10 0;
#X connect 13 0 9 1;
#X connect 13 0 9 0;
#X connect 14 0 5 0;
#X connect 15 0 11 0;
#X connect 16 0 17 0;
#X connect 17 0 0 0;
#X connect 18 0 2 0;
#X restore 546 522 pd synth;
#X floatatom 599 497 4 0 0 0 - - - 0;
#X msg 565 469 0;
#X msg 599 469 90;
#X msg 398 146 mask 4 0.7;
#X text 484 132 Describes how energy in each frequency band masks later energy in the band. Here the masking is total for 4 analysis periods and then drops by 0.7 each period., f 62;
#X text 483 291 Poll the current spectrum via "raw" outlet \, You can set a very high threshold if you don't want attacks mixed in.;
#X msg 471 334 debug 0;
#X obj 464 585 print cooked;
#X obj 385 585 print raw;
#X text 267 555 enable printout:;
#X text 496 216 Minimum "velocity" to output (quieter notes are ignored.), f 59;
#X text 470 53 Set low and high thresholds. Signal growth must exceed the high one and then fall to the low one to make an attack. The unit is the sum of the proportional growth in the 11 filter bands. Proportional growth is essentially the logarithmic time derivative., f 64;
#N canvas 488 108 694 491 creation-arguments 0;
#X text 228 14 creation arguments for bonk~;
#X text 33 114 All frequency parameters are specified in 'bins'. One bin is the sample rate divided by the window size. The minimum possible bandwidth is 1.5 bins. Higher bandwidths give numerically more robust outputs., f 88;
#X text 204 200 window size in points, f 60;
#X text 204 224 analysis period ("hop size") in points, f 60;
#X text 204 247 number of filters to use, f 60;
#X text 204 270 desired bandwidth of filters in halftones \, effective in the exponentially spaced region. (At lower center frequencies the bandwidth is supported by the "minbandwidth" parameter below)., f 60;
#X text 204 385 overlap factor between filters. If 1 \, the filters are spaced to line up at their half-power points. Other values specify more or fewer filters proportionally.;
#X text 33 48 bonk~ uses a filterbank whose center frequencies are spaced equally at low frequencies and proportionally at high ones - i.e. \, they increase linearly \, then exponentially. They are determined by the filters' bandwidths and overlap. The bandwidths are specified proportionally to frequency but bounded below by a specified minimum., f 88;
#X text 204 334 minimum bandwidth in bins. If the bandwidth specified by "halftones" is smaller than this \, this value is used. This must be at least 1.5., f 60;
#X text 204 434 center frequency \, in bins \, of the lowest filter. The others are computed from this., f 60;
#X text 43 173 flags and default values:;
#X text 101 202 -npts (256):;
#X text 108 226 -hop (128):;
#X text 80 249 -nfilters (11):;
#X text 80 271 -halftones (6):;
#X text 45 337 -minbandwidth (1.5):;
#X text 94 388 -overlap (1):;
#X text 87 436 -firstbin (1):;
#X restore 72 530 pd creation-arguments;
#N canvas 654 163 685 300 templates 0;
#X msg 79 97 learn 0;
#X msg 90 137 forget;
#X msg 101 174 write templates.txt;
#X msg 117 204 read templates.txt;
#X msg 52 38 debounce 0;
#X msg 66 66 learn 10;
#X obj 69 253 outlet;
#X text 144 131 Forget the last template. In Learn mode \, use "forget" to erase and record over a template.;
#X text 245 174 Write templates to a file in text-editable format.;
#X text 253 204 Read templates from a file.;
#X text 134 39 Minimum time (msec) between attacks in learn mode;
#X text 130 65 <----;
#X text 169 65 Forget all templates and start learning new ones. The argument gives the number of times you will hit each instrument (10 recommended.) Turn on the output volume above for audible feedback as you train [bonk] - "learn 0" exits learn mode., f 69;
#X connect 0 0 6 0;
#X connect 1 0 6 0;
#X connect 2 0 6 0;
#X connect 3 0 6 0;
#X connect 4 0 6 0;
#X connect 5 0 6 0;
#X restore 501 428 pd templates;
#X msg 371 76 thresh 2.5 5;
#X msg 414 188 attack-frames 1;
#X msg 426 215 minvel 7;
#X msg 434 242 spew 0;
#X msg 444 268 useloudness 0;
#X obj 36 11 bonk~;
#X text 82 10 - attack detection and spectral envelope measurement;
#X text 486 243 Turn spew mode on/off.;
#X text 534 188 Number of frames over which to measure growth., f 54;
#X text 548 267 experimental: use alternative loudness units.;
#X text 532 334 Turn debugging on or off.;
#X text 528 364 Print out settings and templates.;
#X text 552 393 Print out filterbank settings.;
#X msg 738 549 \; pd dsp \$1;
#X text 770 514 DSP on/off;
#X obj 7 42 cnv 1 925 1 empty empty empty 8 12 0 13 #000000 #000000 0;
#X text 840 10 <= click;
#N canvas 585 77 723 566 reference 0;
#X obj 8 35 cnv 5 700 5 empty empty INLET: 8 18 0 13 #202020 #000000 0;
#X obj 8 307 cnv 2 700 2 empty empty OUTLETS: 8 12 0 13 #202020 #000000 0;
#X obj 8 386 cnv 2 700 2 empty empty ARGUMENTS: 8 12 0 13 #202020 #000000 0;
#X obj 7 540 cnv 5 700 5 empty empty empty 8 18 0 13 #202020 #000000 0;
#X obj 7 330 cnv 1 700 1 empty empty 1st: 8 12 0 13 #9f9f9f #000000 0;
#X obj 37 7 bonk~;
#X text 83 6 - attack detection and spectral envelope measurement;
#X obj 7 357 cnv 1 700 1 empty empty 2nd: 8 12 0 13 #9f9f9f #000000 0;
#X obj 7 409 cnv 1 700 1 empty empty flags: 8 12 0 13 #9f9f9f #000000 0;
#X text 158 45 signal - signal input to be analyzed., f 75;
#X text 53 62 thresh <float \, float> - set low and high thresholds., f 90;
#X text 152 413 '-npts': window size in points (default 256)., f 69;
#X text 159 430 '-hop': analysis period ("hop size") in points (default 128)., f 68;
#X text 124 447 '-nfilters': number of filters to use (default 11)., f 73;
#X text 117 464 '-halftones': filter bandwidth of filters in halftones (default 6)., f 74;
#X text 96 481 '-minbandwidth': minimum bandwidth in bins (default 1.5)., f 77;
#X text 131 498 '-overlap': overlap factor between filters (default 1), f 72;
#X text 124 515 '-firstbin': center frequency \, in bins \, of the lowest filter (default 1)., f 73;
#X text 53 96 attack-frames <float> - set number of frames over which to measure growth., f 90;
#X text 88 113 minlevel <float> - set minimum "velocity" to output (ignore quieter notes)., f 85;
#X text 67 147 useloudness <float> - nonzero sets to alternative loudness units (experimental)., f 88;
#X text 172 164 bang - poll the current spectrum via "raw" outlet., f 73;
#X text 109 181 debug <float> - nonzero sets to debugging mode., f 82;
#X text 88 215 debounce <float> - set minimum time (in msec) between attacks in learn mode., f 85;
#X text 109 232 learn <float> - forget and learn for a given number of times (10 recommended)., f 82;
#X text 158 249 forget - forget the last template., f 75;
#X text 109 283 read <symbol> - read templates from a file., f 82;
#X text 102 266 write <symbol> - write templates to a file., f 83;
#X text 109 198 print <float> - print out settings \, templates and filterbank settings for nonzero., f 82;
#X text 67 79 mask <float \, float> - set energy mask profile: number of analysis and drop factor., f 88;
#X text 116 130 spew <float> - nonzero turns spew mode on \, zero sets it off., f 81;
#X text 82 333 list - raw: attack's spectrum (loudness values for the 11 frequency bands used)., f 80;
#X text 82 363 list - cooked: instrument number (if there's a template) \, velocity and temperature., f 83;
#X restore 746 11 pd reference;
#X obj 7 620 cnv 1 925 1 empty empty empty 8 12 0 13 #000000 #000000 0;
#X text 705 631 updated for Pd version 0.42;
#X text 632 467 output volume (0-100);
#X obj 432 483 bonk~;
#X msg 482 365 print;
#X msg 491 393 print 1;
#X obj 367 419 adc~ 1;
#X obj 424 530 tgl 19 0 empty empty empty 17 7 0 10 #dfdfdf #000000 #000000 0 1;
#X obj 503 530 tgl 19 0 empty empty empty 17 7 0 10 #dfdfdf #000000 #000000 0 1;
#X obj 461 296 bng 19 250 50 0 empty empty empty 17 7 0 10 #dfdfdf #000000 #000000;
#X text 15 73 The [bonk~] object takes an audio signal input and looks for "attacks" defined as sharp changes in the spectral envelope of the incoming sound. Optionally \, and less reliably \, you can have [bonk~] check the attack against a collection of stored templates to try to guess which of two or more instruments was hit ([bonk~] is described theoretically in the 1998 ICMC proceedings \, reprinted on msp.ucsd.edu.), f 48;
#X text 15 205 The two outputs of [bonk~] are the raw spectrum of the attack (provided as a list of 11 numbers giving the signal "loudness" in the 11 frequency bands used) \, and the "cooked" output which gives an instrument number (counting up from zero) \, "velocity" and "color temperature". The instrument number is significant only if [bonk~] has a "template set" in memory. The "velocity" is the sum of the square roots of the amplitudes of the bands (normalized so that 100 is an attack of amplitude of about 1). The "temperature" is a sort of 'spectral centroid' that correlates with perceived brilliance., f 53;
#X text 15 382 In this patch \, after starting DSP \, you can print out the raw or cooked output using the two [spigot] objects or listen to a synthesizer output by raising its volume., f 42;
#X text 15 446 By default [bonk~]'s analysis is carried out on a 256-point window (6 msec at 44.1 kHz) and the analysis period is 128 samples. These and other parameters may be overridden using creation arguments as shown in the subpatch below:, f 46;
#X text 596 429 <-- Messages for managing templates.;
#X obj 738 512 ../../doc/5.reference/set-dsp-tgl;
#X connect 0 0 10 0;
#X connect 1 0 11 0;
#X connect 3 0 2 1;
#X connect 4 0 3 0;
#X connect 5 0 3 0;
#X connect 6 0 38 0;
#X connect 9 0 38 0;
#X connect 16 0 38 0;
#X connect 17 0 38 0;
#X connect 18 0 38 0;
#X connect 19 0 38 0;
#X connect 20 0 38 0;
#X connect 21 0 38 0;
#X connect 38 0 1 0;
#X connect 38 1 0 0;
#X connect 38 1 2 0;
#X connect 39 0 38 0;
#X connect 40 0 38 0;
#X connect 41 0 38 0;
#X connect 42 0 1 1;
#X connect 43 0 0 1;
#X connect 44 0 38 0;
#X connect 50 0 30 0;