Cutoff Frequency Chart for 4k

... hold on homie ... there s more numbers in arrive ... ...

however - first it were great if we would discuss those charts - the akai one and the silence ones - and if ya want the joto_o one too - they are all informative imo.

as for the akai and the silence one - atm there s numbers only there which are fukkin different imo - no coments - no conclusions. so please - feel free to criticze the test i did as well as the akai chart - the method and the results of the test ... ie: have i missed out somethin in the process? is the method reliable at all? what about the results? ... and so on ... i will ad some coments later on - but would like to hear someone elses opinions first ... the only aim is to find out more bout those filters!

and yo - blue haze - sorry to ask you again - but if ya ever meet the guys there at akai you would do me a big favour if ya could ask them how they did their test - can you please? ... would be helpful to understand whaz really goin on here ... if so thanx in advance ... and yo - feel free to show them the silence numbers ...

... and yo - some critical coments please ...

... amigos de los filtros ... whatzup!?

First up – if ya aint interested in 4k filters and some possible limitations ya better ignore coz this gets a longwinded post ... i wanted to find out more bout those filters and thanx to elmacaco i found a way how to test them ... so - at startout i just wanted to do one 4lp test and over ... now i got ... dunno - ... too many ... anyways - i did those tests and continue now with some coments on it ... what I mainly was interested was the filters frequency range and whats goin on between that 0 and 20 cutoff ….

in case anyone wants to redo it on his own some more notes on the method: see my post above – i forgot there that the original samples were all set to one shots …, sequence had two midievents on 001:01.000 - one on track 1 to int a 16 and the other on track 2 to the program containing the sample ... and the entire thing has been done on 1.70.

yo - i ll ad now some more filter confusion ... … dunno … see for yourself! ... nah but seriously - to the more knowledgeable users on the matter: please shout out loud if i did anything wrong in the test or if ya think the conclusions are bull - ... the numbers on the akai chart as on the silence test are confusing in itself already and i really dont wanna confuse other 4k heads any further. Imo both of those charts dont make our filters nor any better nor any worse – but hopeffully a bit more knowledgeable – so ... the only aim is to find out more about them. ... besides that – maybe we find out somethin we can tell akai so that they can look into it in future ... (yo – I know it doesnt make sense atm … but anyways ... )

as for adding this to the tips and tricks book – dunno whether that’s a good idea – at the end the akai chart is still somethin official from akai – whilst thisone has been done in a small dark basement … so - I took care to do it quite acurately but still …. dunno … i m no professional.inc .... so - just checkout those numbers and print em out if ya think they are useful for ya is better imo … and if ya want em all – just hit me up - I got em all on excel …


ok - we got some filtertests and charts for our 4k. Currently I know about the akai chart, the joto_o test at yahoo and the silence one – if there are any others around please let me know ... and - since on the tests i had those initial problems with 4ks internal silence i did another one at startout: white noise (c3 (60) at 0db from my virus) resonating over a 4lp and i have another one on normalized silence (but different silence as the one above ... since i had all those silences i liked the one i did it at the end best ... must have felt more silent ... ) – i ll copy those tests all after this post too …

as ever - the ones i posted above have all been done on the same source sample (mono and stereo silence ) and are imo the best reference ... but the others were helpful too imo ... yo - and if ya interested in the matter you should really checkout the filtertest from joto_o at yahoo. That one is dope - he s been makin some very good coments and conclusions to the 4k and its filters there imo.… and - since I include some coments on that one aswell i ll copy his notes here later on too (... yoto hope that’s ok for ya …) … at the end the joto_o test shows 4k filter limitations but imo doesnt tell about freq range and such ... the akai chart and the silence one show limitations too – but show it from another angle imo and maybe they can help to explain some things. if anyone is up to talk about the joto_o one – great! Btw: joto_o s test is quite old (2003) some things might have changed due to new os's ... dunno ... (for example resonance – at the time yoto did it resonance was a 3 digit number)


some coments to the tests:

1.frequency range:

dunno whether anyone noticed - but the chart from akai and the silence ones i did are totally different. so - at least one of the 2 charts must be wrong ... there is no other way around it ... maybe they are even both wrong - who knows! ... main aim for me was to find out the 4k filters freqeuncy range. the silence charts now more than doubles the freq range from the akai one!

Akai chart = 21,5 hz (0) to 6461,5 hz (99!)

Silence test = ca. 47 hz (0) to 15171 hz (100)

... not so easy to determine whichone is more correct here ... but dunno ... 21,5 is fukkin low imo and ... question: do we hear that low at all? I mean 21,5 hz is more than one octave lower than 47 hz – .... and according to what i always thought: ... humans dont really hear at 20 hz ... well dunno – if my cutoff is fully down on a bd or a bass or whateva other low signal at a 2lp, 4lp or 6lp i still hear somethin ... so ... dunno - ... either my ears got really digital over the years or the 47 value is more realistic. as for the higher ranges - .... not sure there either – but again – all the silence tests reach more or less the same value ... and again – take some hh, cy or some other high signal ... and a 6 hp ... dunno – around 15khz sounds more realistic to me too …


2. cutoff progression over the entire range:

the akai chart gives quite a strange progression over the entire range (esp. from 0 to 19 = 21,5 ... and others too). imo this (esp. 0 to 19) is wrong. I got 101 resampled resonating sorta silences here for each filter (and white noises) ... in quite some variations ... if i normalize them and assign them to a program one by one – that program almost plays back as a piano. It has a rather constant progression in pitch if goin up from 0 to 1 to 2 to 10 to 20 to ...100 ... but yo – weird sounds – so i cant really say they go up fully chromatically – but more or less we re there ... (at least its constantly goin upwards)

So imo the cutoff progression is quite constant over the entire range: esp. regarding the 2lp. but even the 4lp and the 6lp are quite constant – the weird numbers on the 4lp and 6lp are imo explained due to other things ...


3. cutoff progression in the lower ranges:

one of the little things that both tests have in common is that they show that somethin in the lower freq ranges is not ok. (joto_o s test shows that even better imo – its audible there) akai gives 21,5 hz from cutoff - 0 to 19. silence test gives quite smooth values for the 2lp - but than the 4lp and esp. the 6lp are quite freakin out on certain cutoff settings. so somethin here is definately strange ....

well – again - i got all those 101s of resonating silence samples and i looked a bit further into them. basically – initially they more or less all look the same – esp. in the lower ranges – its resonating silence – so there s almost nothin to see there. but if ya normalize the resonating silences they almost look like a sinewave – with little cycles in the low ranges and lotsof in the highs - … ya can count the cycles easily for the 0,1,2 values ... (I counted some to proove whether the waveeditor displays cerocrossings correctly – they do!) … as ever - i rechecked some or most of the freakin out values on the 6lp (ie: cutoff 0,1, and some others up til 20) and on the 4lp. those freakin out samples (values) all show more or less the same thing: the respective freakin out parts of the wave (left or right – it is not always the same side) have a similar waveform (for example more or less same amount of cycles on 2lp, 4lp, 6lp at 0 which is 47) but some waves (or parts of it) are not crossing the centerline (baseline or infinity). this phenomenon is called dc offset (explanation for it comes later)

Besides that some waves show irregularities in the cycles … sometimes the left side does 5 cycles whilst the right one does 1 only (for example Silence 4lp Stereo at cutoffsetting 2!) – … and its not really a normal cycle as on the others but a weird line – its not really lookin as a sine on those fragments of sound at all!


Dc offset: this is what soundforge and wavelab say on the matter …

soundforge:

DC offset occurs when hardware, such as a sound card, adds DC current to a recorded audio signal. This current results in a recorded waveform that is not centered around the baseline (-infinity). Glitches and other unexpected results can occur when sound effects are applied to files that contain DC offsets. Sound Forge can compensate for this DC offset by adding a constant value to the samples in the sound file.

In the following example, the red line represents the baseline. The lower waveform exhibits DC offset; note that the waveform is centered approximately 2 dB above the baseline.

(My coment: sorry – I cant post the pic here that soundforge added - but if ya have soundforge – hit f1 - search for dc offset - and you ll see what it is grafically – and its really almost the same on the freakin out resonating silences as there – just that on the soundforge example dc offset is positive (over the infinity line) – in our freakin 6lp and 4lp parts there is a negative dc offset (under the infinity line) by some db.)



this is what wavelab says on the matter:

A DC offset is when there is too large a DC (direct current) component in the signal. If the DC offset is really bad, it can be visible as the signal not being visually centered around the "zero level axis". However, the DC offset can be significant without actually being seen.

A DC offset is problematic for two reasons:
It affects where the zero crossings appear, which in turn affects the smoothness of splices between audio files.
Certain processing options do not give optimal results when performed on files with a DC offset.



So – for the curoff progression in the lower ranges I think that the weird numbers in the silence test (esp. 6lp, 4lp) are due to 1. dc offset and 2. those other irregularities in the waves – whilst the cutoff progression is quite progressive – even in the lower ranges (both – 4lp and 6lp do ca. 47 cycles at for example cutoff 0 – just that all (6lp at 0 and 1) or most of the cycles are under the infinity line.)

That dc offset looks quite scary – and made me do some more tests on those lower ranges – freaked some full level signals into those filtes: sine, saw and white noise from my virus – and on all of them there s no dc offset to see … also if I play them back they do not sound unpredictable too (as the silence ones do indeed) – again – hard to tell bout chromatically – but more or less its ok – so imo dc offset happens on extreme samples and extreme settings of resonance … (also yoto added silence in his test to the impulse sample – guess that could be a reason why of some of the weird sounds there …. ) … so dunno here – but on 0db signals this doesn’t seem to be a big problem …. so – dunno - its just somethin i noticed whilst doin the test and i thought I let y’all know … hope someone else can coment more on thisone …. it is surely not 100% ok imo but at least it helped me to read those silence charts better …


4. mono vs. Stereosamples ...

at startout i started with stereo silence samples ... this way at times there are different hz values for left and right ... yo – on mono samples this is not the case – so the differences between left and right must be due to the different input signal ... (only question I got here – why is my right in more silent as the left one? - yo I know its almost nada – but still …)


5. different hz values on 0db samples ...

havent done the entire thing (2lp,4lp,6lp) on sine, saw, white noise and normalized silence (just 0-20 and 90-100). i mainly did them to checkout more on the dc offset. ... the hz values on those tests differ from the silence ones the further goin up the freq range – but thats in the nature of the original sample and logic imo ... (used c3 sines and saws and white noise) ... i ll add the hz values for the original samples on every test – and befor watchin the hz chatts – checkout those values ... on the other side the hz values in the 0,1,2,3,5,10 cutoff range – are almost the same on those tests as on the silence tests - which quite confirms the values for the silence tests in the lower ranges ...


so dunno - i m quite convinced now that those filtes range from 47 to 15171 and that the filtesrs are progressive ... (best reference for me atm is silence mono 2lp) ... i dont really know what to say on the dc offset but i hope someone else jumps in here ... as i see it atm its only apparent on extreme settings ... (...and seems bein only there for 4lp and 6lp ) ... test is over – lookin forward to upload 1.71 now ... and yo – feel free to see things differently .... – yo – many thanx again to elmacaco – ... and coments are as always very welcome ...

Original Sample = SILENCE NORMALIZED (L = 33923,68, R = 34008,15)



0 47 46 46,5
1 43 47 45
2 52 52 52
3 53 54 53,5
4 54 56 55
5 59 59 59
6 63 57 60
7 69 68 68,5
8 74 72 73
9 76 78 77
10 77 82 79,5
11 82 87 84,5
12 91 92 91,5
13 95 99 97
14 105 104 104,5
15 110 108 109
16 114 114 114
17 117 118 117,5
18 129 128 128,5
19 138 137 137,5
20 149 147 148
21 158 156 157
22 165 167 166
23 179 177 178
24 188 187 187,5
25 198 198 198
26 204 207 205,5
27 215 214 214,5
28 217 223 220
29 239 238 238,5
30 256 256 256
31 274 273 273,5
32 293 294 293,5
33 315 315 315
34 333 330 331,5
35 357 356 356,5
36 376 366 371
37 394 391 392,5
38 415 413 414
39 435 429 432
40 454 452 453
41 474 463 468,5
42 513 512 512,5
43 551 549 550
44 595 594 594,5
45 628 628 628
46 666 665 665,5
47 712 715 713,5
48 741 753 747
49 790 791 790,5
50 830 834 832
51 870 869 869,5
52 907 908 907,5
53 946 943 944,5
54 1028 1025 1026,5
55 1102 1097 1099,5
56 1176 1172 1174
57 1265 1266 1265,5
58 1342 1329 1335,5
59 1415 1411 1413
60 1506 1501 1503,5
61 1581 1574 1577,5
62 1654 1651 1652,5
63 1746 1740 1743
64 1820 1820 1820
65 1894 1893 1893,5
66 2058 2059 2058,5
67 2207 2209 2208
68 2357 2358 2357,5
69 2536 2538 2537
70 2686 2681 2683,5
71 2832 2835 2833,5
72 3014 3012 3013
73 3163 3162 3162,5
74 3311 3315 3313
75 3496 3497 3496,5
76 3647 3645 3646
77 3798 3797 3797,5
78 4067 4068 4067,5
79 4429 4421 4425
80 4730 4732 4731
81 5033 5031 5032
82 5394 5396 5395
83 5699 5699 5699
84 5998 6001 5999,5
85 6365 6370 6367,5
86 6672 6675 6673,5
87 6980 6981 6980,5
88 7347 7347 7347
89 7655 7657 7656
90 8209 8212 8210,5
91 8954 8956 8955
92 9579 9581 9580
93 10209 10213 10211
94 10968 10972 10970
95 11610 11611 11610,5
96 12255 12258 12256,5
97 13040 13039 13039,5
98 13693 13695 13694
99 14364 14361 14362,5
100 15170 15173 15171,5

Original Sample: White Noise (L = 15992,88, R = 16232,76)



0 47 47 47
1 49 49 49
2 52 52 52
3 55 55 55
4 58 58 58
5 59 59 59
6 64 64 64
7 69 69 69
8 74 74 74
9 78 79 78,5
10 85 85 85
11 89 89 89
12 93 93 93
13 100 100 100
14 105 105 105
15 109 109 109
16 114 114 114
17 120 120 120
18 128 128 128
19 140 140 140
20 148 148 148
21 159 159 159
22 168 168 168
23 180 181 180,5
24 188 188 188
25 199 199 199
26 207 207 207
27 216 216 216
28 227 227 227
29 239 239 239
30 257 257 257
31 276 276 276
32 299 299 299
33 316 316 316
34 334 334 334
35 358 358 358
36 376 376 376
37 395 395 395
38 418 418 418
39 436 436 436
40 455 455 455
41 477 477 477
42 514 514 514
43 552 552 552
44 596 596 596
45 632 632 632
46 670 670 670
47 715 715 715
48 754 754 754
49 790 790 790
50 835 835 835
51 872 872 872
52 910 910 910
53 946 946 946
54 1028 1028 1028
55 1103 1104 1103,5
56 1178 1178 1178
57 1268 1268 1268
58 1343 1343 1343
59 1419 1419 1419
60 1505 1507 1506
61 1582 1582 1582
62 1656 1656 1656
63 1746 1745 1745,5
64 1819 1820 1819,5
65 1896 1896 1896
66 2059 2059 2059
67 2209 2209 2209
68 2360 2358 2359
69 2535 2536 2535,5
70 2682 2684 2683
71 2837 2841 2839
72 3013 3017 3015
73 3164 3161 3162,5
74 3309 3310 3309,5
75 3490 3495 3492,5
76 3639 3642 3640,5
77 3790 3793 3791,5
78 4050 4050 4050
79 4415 4420 4417,5
80 4703 4710 4706,5
81 5002 5005 5003,5
82 5353 5362 5357,5
83 5645 5658 5651,5
84 5942 5957 5949,5
85 6296 6306 6301
86 6573 6575 6574
87 6883 6896 6889,5
88 7230 7238 7234
89 7499 7543 7521
90 8040 8043 8041,5
91 8748 8775 8761,5
92 9303 9358 9330,5
93 9887 9955 9921
94 10554 10635 10594,5
95 11116 11211 11163,5
96 11718 11800 11759
97 12411 12508 12459,5
98 12943 13083 13013
99 13554 13697 13625,5
100 14251 14419 14335

Original Sample: SINE L (L = 261,78 ) = c3 (60)


0 45 45 45
1 47 47 47
2 50 50 50
3 51 51 51
4 54 54 54
5 56 56 56
6 60 60 60
7 64 64 64
8 68 68 68
9 71 71 71
10 76 76 76
11 80 80 80
12 88 88 88
13 88 88 88
14 88 88 88
15 89 89 89
16 95 95 95
17 98 98 98
18 104 104 104
19 111 111 111
20 156 156 156

...

90 265 265 265
91 277 277 277
92 268 268 268
93 276 276 276
94 302 302 302
95 285 285 285
96 343 343 343
97 315 315 315
98 416 416 416
99 364 364 364
100 514 514 514

0 48 48 48
1 50 50 50
2 52 52 52
3 54 54 54
4 57 57 57
5 60 60 60
6 64 64 64
7 70 70 70
8 75 75 75
9 79 79 79
10 85 85 85
11 90 90 90
12 94 94 94
13 100 100 100
14 105 105 105
15 109 109 109
16 115 115 115
17 120 120 120
18 129 129 129
19 140 140 140
20 149 149 149

...

90 262 262 262
91 268 268 268
92 262 262 262
93 266 266 266
94 310 310 310
95 277 277 277
96 315 315 315
97 288 288 288
98 394 394 394
99 343 343 343
100 513 513 513

0 47 47 47
1 50 50 50
2 52 52 52
3 54 54 54
4 57 57 57
5 60 60 60
6 64 64 64
7 70 70 70
8 75 75 75
9 79 79 79
10 84 84 84
11 89 89 89
12 94 94 94
13 99 99 99
14 104 104 104
15 109 109 109
16 114 114 114
17 119 119 119
18 128 128 128
19 139 139 139
20 149 149 149

...

90 262 262 262
91 262 262 262
92 262 262 262
93 262 262 262
94 272 272 272
95 262 262 262
96 291 291 291
97 269 269 269
98 361 361 361
99 329 329 329
100 512 512 512

Original Sample = SAW (L 262) = c3 60



0 53 53 53
1 58 58 58
2 59 59 59
3 59 59 59
4 62 62 62
5 65 65 65
6 69 69 69
7 75 75 75
8 80 80 80
9 87 87 87
10 93 93 93
11 97 97 97
12 101 101 101
13 110 110 110
14 115 115 115
15 116 116 116
16 123 123 123
17 130 130 130
18 136 136 136
19 150 150 150
20 160 160 160

...

90 4395 4395 4395
91 4682 4682 4682
92 5041 5041 5041
93 5533 5533 5533
94 5904 5904 5904
95 6213 6213 6213
96 6268 6268 6268
97 6482 6482 6482
98 7050 7050 7050
99 6821 6821 6821
100 7107 7107 7107

0 48 48 48
1 50 50 50
2 52 52 52
3 54 54 54
4 57 57 57
5 60 60 60
6 64 64 64
7 70 70 70
8 75 75 75
9 79 79 79
10 85 85 85
11 90 90 90
12 94 94 94
13 100 100 100
14 105 105 105
15 109 109 109
16 115 115 115
17 120 120 120
18 129 129 129
19 140 140 140
20 149 149 149

...

90 4377 4377 4377
91 4621 4621 4621
92 4848 4848 4848
93 5414 5414 5414
94 5605 5605 5605
95 6125 6125 6125
96 6110 6110 6110
97 6312 6312 6312
98 6935 6935 6935
99 6815 6815 6815
100 7207 7207 7207

SAW 6 LP L SAW 6 LP R SAW 6 LP
0 47 47 47
1 50 50 50
2 52 52 52
3 54 54 54
4 57 57 57
5 60 60 60
6 64 64 64
7 70 70 70
8 75 75 75
9 79 79 79
10 84 84 84
11 90 90 90
12 94 94 94
13 99 99 99
14 104 104 104
15 109 109 109
16 114 114 114
17 119 119 119
18 128 128 128
19 139 139 139
20 149 149 149

...

90 4347 4347 4347
91 4655 4655 4655
92 4714 4714 4714
93 5253 5253 5253
94 5446 5446 5446
95 5841 5841 5841
96 6006 6006 6006
97 6247 6247 6247
98 6912 6912 6912
99 6889 6889 6889
100 7307 7307 7307

check it out there ...

http://launch.groups.yahoo.com/group/mpc4000/

joto_0's coments:

This patch demonstrates the use of an impulse-sample ('dirac') to excite the filters and use a sampler for rudimentary physically-inspired modelling.

observations:

* "Funny" tuning scale of the filters in the middle octaves
The filter resonance frequencies are very roughly quantized, not exactly chromatic scale. This is not a user setting.
The control over the filter resonance has a very rough (100 units!?) resolution too. It is not possible to use smaller increases in resonance freq's than about a semitone, even when using the modulation matrix.
In this patch the alternative tuning sounds cool, but it's still a BUG.

* the filters break up at high-pitches
TAKE CARE FOR YOUR EARS.
Analog-synth-freaks say digital is shit. In this case they ARE right. But it's a horrible implementation, not an inherent defect in digital synthesis.

* If the impulse-sample wave starts with the pulse, the sampler is silent. Does the Akai skip the first sample?

* The loop in the wave (on silence) is needed. Otherwise the engine decides to cut the voice while the filters are still ringing.



This patch is made on a Z8.

Well yes,
although we've known all this since 2003!
Akai Unprofessional Inc never deigned to comment on it though.

feline1 wrote:Well yes,
although we've known all this since 2003!
Akai Unprofessional Inc never deigned to comment on it though.

well ... ... dunno feline ... i know off a chart someone there at akai just handed out about 2 month ago - and i doubt that one single number on it is correct ... or is there another freq range around - if so could you please point me to it? ... and - a few pages back you talked about a freq response that copmletely sucks ... imo it aint that bad ...

j0t0's researches from 2+ years ago are correct (you can HEAR them!
no arguing with your ears)

Akai's chart is utter made-up fantasy.