Another EQ this time only with peak biquad and only dealing with peaks (and no valley) on the freq curve.
Algorithm try to flatten the LW and the Early Reflections.
The table below gives the impact on the pref score with a growing number of PEQs:
Code:
INFO:spinorama:OPTIM JBL 708P START ['Listening Window', 'Early Reflections'] #PEQ 20 Freq #5 Gain #5 +/-[0.2, 12] Q #5 [0.1, 12] Loss 722.43 Score 4.97
INFO:spinorama:Speaker JBL 708P Iter 0 Optim converged loss 10.11 pref score 5.08 biquad PK F: 911Hz Q:6.95 G:-3.24dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 1 Optim converged loss 5.52 pref score 5.61 biquad PK F: 9012Hz Q:0.40 G:-2.15dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 2 Optim converged loss 8.70 pref score 5.69 biquad PK F:14259Hz Q:6.62 G:-2.25dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 3 Optim converged loss 8.84 pref score 5.74 biquad PK F: 669Hz Q:12.00 G:-1.13dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 4 Optim converged loss 8.75 pref score 5.75 biquad PK F: 370Hz Q:12.00 G:-0.67dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 5 Optim converged loss 8.56 pref score 5.84 biquad PK F: 3660Hz Q:12.00 G:-1.03dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 6 Optim converged loss 8.59 pref score 5.88 biquad PK F: 3010Hz Q:12.00 G:-0.75dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 7 Optim converged loss 8.66 pref score 5.92 biquad PK F: 2549Hz Q:12.00 G:-0.71dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 8 Optim converged loss 8.64 pref score 5.96 biquad PK F: 7105Hz Q:12.00 G:-0.59dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 9 Optim converged loss 8.68 pref score 5.97 biquad PK F: 987Hz Q:12.00 G:-0.38dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 10 Optim converged loss 8.65 pref score 5.99 biquad PK F: 5326Hz Q:12.00 G:-0.27dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 11 Optim converged loss 8.78 pref score 5.96 biquad PK F: 527Hz Q:12.00 G:-0.74dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 12 Optim converged loss 8.84 pref score 5.99 biquad PK F: 671Hz Q:12.00 G:-0.71dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 13 Optim converged loss 8.82 pref score 6.00 biquad PK F: 1766Hz Q:12.00 G:-0.44dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 14 Optim converged loss 8.94 pref score 5.97 biquad PK F: 797Hz Q:12.00 G:-0.49dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 15 Optim converged loss 9.03 pref score 5.93 biquad PK F: 510Hz Q:12.00 G:-0.47dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 16 Optim converged loss 9.10 pref score 5.95 biquad PK F: 669Hz Q:12.00 G:-0.46dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 17 Optim converged loss 9.07 pref score 5.96 biquad PK F: 1334Hz Q:12.00 G:-0.25dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 18 Optim converged loss 9.15 pref score 5.93 biquad PK F: 797Hz Q:12.00 G:-0.31dB in 500 iterations
INFO:spinorama:Speaker JBL 708P Iter 19 Optim converged loss 9.22 pref score 5.91 biquad PK F: 533Hz Q:12.00 G:-0.30dB in 500 iterations
INFO:spinorama:OPTIM END JBL 708P: best loss 9.16 final score 5.91 with 20 PEQs
Here is the full EQ. I would keep only the first 3 after that it is mostly sharp filters to remove small oscillations that may or may not be audible.
Code:
Preference Score 5.0 with EQ 5.9
Generated from http://github.com/pierreaubert/spinorama/generate_peqs.py v0.5
Dated: 2021-02-07-16:44:53
Preamp: -0.2 dB
Filter 1: ON PK Fc 910 Hz Gain -3.24 dB Q 6.95
Filter 2: ON PK Fc 9012 Hz Gain -2.15 dB Q 0.40
Filter 3: ON PK Fc 14259 Hz Gain -2.25 dB Q 6.62
Filter 4: ON PK Fc 668 Hz Gain -1.13 dB Q 12.00
Filter 5: ON PK Fc 369 Hz Gain -0.67 dB Q 12.00
Filter 6: ON PK Fc 3660 Hz Gain -1.03 dB Q 12.00
Filter 7: ON PK Fc 3010 Hz Gain -0.75 dB Q 12.00
Filter 8: ON PK Fc 2549 Hz Gain -0.71 dB Q 12.00
Filter 9: ON PK Fc 7105 Hz Gain -0.59 dB Q 12.00
Filter 10: ON PK Fc 987 Hz Gain -0.38 dB Q 12.00
Filter 11: ON PK Fc 5326 Hz Gain -0.27 dB Q 12.00
Filter 12: ON PK Fc 527 Hz Gain -0.74 dB Q 12.00
Filter 13: ON PK Fc 670 Hz Gain -0.71 dB Q 12.00
Filter 14: ON PK Fc 1766 Hz Gain -0.44 dB Q 12.00
Filter 15: ON PK Fc 796 Hz Gain -0.49 dB Q 12.00
Filter 16: ON PK Fc 509 Hz Gain -0.47 dB Q 12.00
Filter 17: ON PK Fc 669 Hz Gain -0.46 dB Q 12.00
Filter 18: ON PK Fc 1333 Hz Gain -0.25 dB Q 12.00
Filter 19: ON PK Fc 796 Hz Gain -0.31 dB Q 12.00
Filter 20: ON PK Fc 533 Hz Gain -0.30 dB Q 12.00
You will notice that the complete filter (the sum of all the biquad, picture left, down) doesn't show sharp peak but a relative smooth curve.
The manual filter is from flipflop@.
View attachment 111061
You can look at the impact on the spinorama:
View attachment 111062
and on the PIR:
View attachment 111063