AutoEq (https://github.com/jaakkopasanen/AutoEq) (by Jaakko Pasanen) is an utility for headphone equalization.
I made routines to make it support SuperEQ, which can be found on few software as like Foobar2k DT/Foobar Mobile and DeaDBeeF (originally SuperEQ was made for Winamp by N. Shibata). At this point AutoEq has to be installed in your system with these routines ... . Here are the functions to be added in
frequency_response.py :
Aitken interpolation (you can use some interpolation method from library as well):
def aitkeninterp(x, y, xi):
n = np.size(x)
if n != np.size(y):
return
A = np.zeros((n,n), float)
D = np.zeros(n, float)
yi = 0.0
for i in range(0, n):
D[i] = xi - x[i]
A[i, 0] = y[i]
for i in range(1, n):
for j in range(1, i+1):
A[i, j] = (D[j-1] * A[i,j-1] - D[i] * A[j-1,j-1]) / (x[i] - x[j-1]);
yi = A[n-1, n-1];
return round(yi)
SuperEQ preset file calculator/writer (not sure of the file extension 'seq'):
def write_supereq(self, file_path):
"""Writes SuperEq's 18 band eq settings to a preset file."""
"""Gain values are interpolated using Aitken's interpolation method from same memory data which is stored into csv file by AutoEQ"""
last_idx = np.size(self.frequency)-1
prev_idx = 0
with open(file_path.replace('.txt', '.feq'), 'w', encoding='utf-8') as f: # .feq = foobar2k preset file extension
s = ''
file_path = os.path.abspath(file_path)
cfreq = [55.0, 77.0, 110.0, 156.0, 220.0, 311.0, 440.0, 622.0, 880.0, 1200.0, 1800.0, \
2500.0, 3500.0, 5000.0, 7000.0, 10000.0, 14000.0, 20000.0] # SuperEQ band center frequencies
idx = 0
for i in cfreq:
idx = bisect.bisect_left(self.frequency, i, lo=prev_idx, hi=last_idx-1)
if last_idx > idx+1:
G = FrequencyResponse.aitkeninterp([self.frequency[idx-2], self.frequency[idx-1], self.frequency[idx], self.frequency[idx+1]], \
[self.equalization[idx-2], self.equalization[idx-1], self.equalization[idx], self.equalization[idx+1]], i)
s += f'{G:d}\n'
prev_idx = idx
else: # get value from the last element
G = round(self.equalization[-1])
s += f'{G:d}\n'
f.write(s)
! remember
import bisect in
frequency_response.py header.
I have tested functionality by calling the function from
write_eqapo_parametric_eq() with a command line
FrequencyResponse.write_supereq(self, file_path).
At least this allows generation of presets for SuperEQ/Foobar2k/etc. without need for full SuperEQ implementation.
As for an example, for AKG K240 Studio I got this data for the preset : 6, 4, 1, -2, -3, -3, -2, -1, -1, 1, 4, -1, 3, 3, -1, -1, 1, 5
Compare : https://github.com/jaakkopasanen/AutoEq/tree/master/results/oratory1990/over-ear/AKG%20K240%20Studio
Code changes needed to support SuperEQ preset file generation directly with a commandline parameter --supereq :
__main__.py
38 arg_parser.add_argument('--supereq', action='store_true',
help='Will produce SuperEQ settings if this parameter exists, no value needed.') batch_processing.py
19 def batch_processing( .... ,supereq=False, ...)
33 if not compensation and (... or supereq):
112 args = (... ,supereq, ...)
133 def process_file(..., supereq, ...)
163 if supereq:
supereq = True
233 if supereq:
fr.write_supereq(output_file_path.replace('.csv', '.feq'))
frequency_response.py
38 def __init__(..., supereq=None, ...)
67 self.supereq = self._init_data(supereq)
73 def copy(..., supereq=self._init_data(self.supereq), ...)
def reset(..., supereq=True, ...)
160 if supereq:
self.supereq = self._init_data(None)
907 def compensate(...
self.reset(..., supereq=True, ...)
)
1044 def smoothen_fractional_octave(...
self.reset(..., supereq=True, ...)
)
Tested with console command:
$ python3 -m autoeq --input-file="measurements/oratory1990/data/over-ear/AKG K240 Studio.csv" --output-dir="my_results" --compensation="compensation/harman_over-ear_2018_wo_bass.csv" --supereq
I was able to make it work by switching back to the version of AutoEq this additional code was most likely written for:
git checkout 402ccd4c0497e70a01263a9a77691535de6f8b8c
Thanks for this by the way! heehee (and sorry for necroposting)