In need of a tutorial on using the Graphical EQ component.

2013-06-20 12:00:20

Hi everyone.

Before now, I always used or customised the EQ to what, to my ears, sounds good.

From what I've read, the true purpose of an EQ is to make the response as flat as possible. Don't have the source of that claim, but this is something I've seen a couple of times.

Quite frankly, how don't know how to do this at all.

Anyone got a well detailed tutorial ?

I already saw http://www.head-fi.org/t/413900/how-to-equ...ones-a-tutorial but it seems that some a the links to the tools used are dead.

What are your methods, programs ?

My headphones are the Sennheiser HD 449.


Thanks

Alan

In need of a tutorial on using the Graphical EQ component.

Reply #1 – 2013-06-21 02:00:27

Ok, so I've looked at google some more, and here's what I know.

Pink noise seems to be the most used noises for EQing.

Also, I've seen this http://www.recordingeq.com/Subscribe/tip/tascam.htm , which helps a lot to know which frequency does what.

Here the thing that I'm trying to do:

I want to create some pink noises at different frequencies, play them in foorbar2000 with my EQ, and have something (a VST ?) that would show a graphic of the frequency response so I could see if the response is flat enough or if I'm way out of it.


Thank you for your help

Alan

In need of a tutorial on using the Graphical EQ component.

Reply #2 – 2013-06-21 04:20:58

Sorry, but I can't edit my post.

Here's what I did.

- Created a 30 seconds pink noise in Audacity

- Converted it to stereo

- Played it in foobar2000 with the EQ as the only DSP, except the next one

- Used a Spectrum analyzer VST

- The DSP Window looks like this
- Equalizer
- Spectrum Analyzer VST

- Played the pink noise and messed with the EQ bands until the response looked like this: http://puu.sh/3kzVb.jpg

- Saved the EQ setting. When auto leveled, it looks like this: http://puu.sh/3kAll.jpg


So, does that makes sense to you ? Am I out of the track ?

Thanks !

Alan

In need of a tutorial on using the Graphical EQ component.

Reply #3 – 2013-06-21 07:17:15

Hi Aldem,
you seem to be missing a crucial point: the purpose of equalizing isn't to correct foobar2000s output (which, by itself, is perfect), but rather to adjust the frequency response of your whole playback chain, especially the headphones.
Therefore, you'd have to measure the output of your headphones using a microphone, which can be rather involved, since the mic has its own freq.response and other aspects like reverberation might affect the result.
Alternatively you could look at the frequency response of your headphones (this can be found on the internet) and do the adjustments accordingly.
That being said, setting the EQ so it sounds good to you, certainly isn't a bad approach either, since music, in the end, is a matter of personal taste and not just physical measurements.
By the way, pink noise is supposed to look like this: http://upload.wikimedia.org/wikipedia/comm...se_spectrum.svg , i.e. a falling curve with constant slope. What you were aiming at, is white noise.

In need of a tutorial on using the Graphical EQ component.

Reply #4 – 2013-06-21 09:57:03

Because this is more directed towards general concepts of EQing rather than anything specific to foo_dsp_xgeq, I’m moving it to General Audio in the main forum, where I think it’ll also gain more attention.

In need of a tutorial on using the Graphical EQ component.

Reply #5 – 2013-06-21 10:18:01

I only use EQ to try and correct obvious deficiencies in the output, regardless of whether it's caused by the hardware or the file.

At home, I don't use EQ at all. At work, I have someone's throwaway cans (Philips SHP2700) plugged straight into the front port of a cheap onboard soundcard (a Dell Vostro 420), and although I have no idea which part of the chain causes it, I experience strong peaks in the <100Hz and ~1KHz area, for which I have to compensate.

Without EQ, it's like the cans sound like actual tin cans, plus a subwoofer (which are clearly invented only to annoy other people) Work is a very noisy environment, compared to home, so any hope of maximum fidelity must be abandoned anyway.

In need of a tutorial on using the Graphical EQ component.

Reply #6 – 2013-06-21 20:19:34

I tipically use EQ to adjust the sound of the output, which more often than not are headphones, to my preference. I don't use curves, just listen and compensate whichever frequencies I find too quiet or too loud, aiming for an overall more "balanced" sound. Pretty similar, I guess, to what you are aiming for with the process you described.

However, bear in mind that with headphones it's not always correct to aim to a completely flat curve, as resonances in the inner ear have to be accounted for (maybe someone with more knowledge could clarify that?). That's why you sometimes see bumps around the 7000hz mark in headphone response graphs.

You can also look for headphones that sound good to you without the need for further EQing. Perhaps in your area there are stores that let you try before buying, if not it's pretty much a guessing game I'm afraid...

In need of a tutorial on using the Graphical EQ component.

Reply #7 – 2013-06-22 03:41:20

You could try using a sweep test, and adjust the EQ so that you hear each tone at a comparable strength.

You could use tests 24-61 here:
http://www.wavecor.co.uk/testdisk.html

It has sine-wave tones recorded at different frequencies, and a different file for each tone. So unlike sweep tests, you know exactly what tone is being played, and can adjust the EQ for that tone.

Note: The tests are in FLAC (lossless), if you need to convert them to a lossy codec, I suggest you take the needed precautions to preserve the highest quality possible, and avoid high-pass or low-pass filters.

In need of a tutorial on using the Graphical EQ component.

Reply #8 – 2013-06-22 22:15:05

Isn't Foobar's default EQ a linear phase (FIR) affair that creates nice steps in the frequency response (with the corresponding issues in time domain)? (It was an adaptation of the Shibatch Super EQ, if memory serves.) That's the wrong approach to EQ entirely. I stopped using it since when the steps exceeded a few dB, sound quality suffered to these ears (and have actually not needed to use an EQ at home for some time). Transducers are plain ol' minimum phase systems and should be EQ'd using an IIR algorithm. Parametric EQs usually are. The general EQing approach obviously is still valid.

dhromed, the ~100 Hz peak is likely to be a result of high output resistance (the usual driver impedance interaction), while the ~1 kHz peak presumably is ordinary earpad- or enclosure-related "honk". If construction is anything like my SHP805s of yore, there may be a good amount of hollow emptiness behind the drivers that sound can bounce around in.