Octatrack Equalizer: Filter Frequency Mapping

I’ve been doing some empirical work to figure out what the FRQ parameter actually means in Hz for the Equalizer’s shelf filters. Both follow an approximately exponential frequency mapping — about 10 parameter units per octave across most of the range, with some nonlinearity at the extremes.

A couple of things worth noting: the Q parameter has no effect on shelf filter response. And the default value of 63 corresponds to ~80 Hz on the high shelf and ~1200 Hz on the low shelf.

Lookup tables below, computed from fits to the measured data. Use them as a starting point — ears still win. Full methodology and plots in the attached writeup.

High shelf

Hz (approx) FRQ
20 43
25 46
31.5 49
40 52
50 55
63 58
80 62
100 66
125 70
160 74
200 78
250 82
315 86
400 89
500 93
630 96
800 100
1k 102
1.25k 105
1.6k 108
2k 110
2.5k 113
3.15k 115
4k 117
5k 119
6.3k 121
8k 123
10k 125
12.5k 126

Values above 5k are less reliable.

Low shelf

Hz (approx) FRQ
25 4
31.5 8
40 12
50 15
63 19
80 23
100 26
125 29
160 33
200 36
250 40
315 43
400 46
500 50
630 53
800 57
1k 60
1.25k 63
1.6k 67
2k 71
2.5k 74
3.15k 78
4k 82
5k 86
6.3k 90
8k 95
10k 99
12.5k 104
16k 111
20k 117

OT_EQ_Frequency_Mapping.docx (66.9 KB)

12 Likes

Following up on the shelf filter measurements, I’ve done the same characterization for the parametric mid EQ.

FRQ mapping

Clean exponential (linear in log space) across the full range:

Hz = 2^(2.3866 + 0.0933 × FRQ)

R² = 0.9996. About 10.7 FRQ units per octave throughout. The default FRQ = 63 lands at ~310 Hz, and the model predicts FRQ = 127 reaches ~19,300 Hz — consistent with the range being designed to top out at 20 kHz. Simpler than the shelf filters, which needed cubic fits.

Q mapping

Also a clean exponential:

Q = 2.0 × 13.0^(param / 127)

Q = 2.0 at param 0, Q = 7.3 at the default (param 63), Q = 26.0 at param 127. Verified at five points with residuals under 0.1 throughout.

Lookup table

Hz FRQ Hz (predicted)
20 21 20.3
25 24 24.7
31.5 28 32.0
40 31 38.8
50 35 50.3
63 38 61.0
80 42 79.0
100 46 102.3
125 49 124.2
160 53 160.9
200 56 195.3
250 60 253.0
315 63 307.1
400 67 397.7
500 71 515.1
630 74 625.4
800 78 809.9
1000 81 983.3
1250 85 1273.5
1600 89 1649.3
2000 92 2002.3
2500 95 2430.8
3150 99 3148.2
4000 103 4077.3
5000 106 4949.9
6300 110 6410.7
8000 113 7782.8
10000 117 10079.6
12500 120 12237.0
20000 127 19291.8

Full writeup with methods and figure attached.

OT_Mid_EQ_Mapping.docx (71.1 KB)

7 Likes

Thanks @Bryan_T !

Would be great if you could do it for Filter FX and with tuning and fine tuning like for Lofi AMF. The idea would be to use it as a tuned band pass resonator…

How do you reverse engineered formulas ?

I’ll tackle the Filter next. These examples were to work out a method without adding the additional fine-tuning parameter. Now that I have a pretty quick method, the Filter isn’t too daunting. As long as Logic has a filter with similar behavior.

The formulas are a regression model that I have Claude run. Basically, I had my fifteen data points that phase cancelled really well, plotted the relationship, noted it was basically linear if I transformed the data (not a surprise for an EQ), then asked Claude to run the regression.

3 Likes

Famous last words! The approach I used for the Equalizer does not nearly as well for the Filter. I’ve got some other ideas to explore.

4 Likes