[PEQ game] If you only had 3 (or 4) filters...

[Jimbob54]

That's correct. A sharp cut within the audible frequency range can cause audible "ringing." A steeper slope also requires a longer filter, so there will be some limit to what is possible with a given processor. Modern DSPs are fast enough that that's not likely to be the limiting factor, though.

I might have learnt something in my couple of years here. Thanks to your good self and many others with actual knowledge!

Now I just need to figure out what ringing is and how to discern it

 

[Jose Hidalgo]

A sharp cut within the audible frequency range can cause audible "ringing."

You haven't addressed this though :

On the other hand, if such cut-offs are at both ends of the spectrum (20-30 and 18K-20K), wouldn't the ringing become hard to hear ?

 

[mansr]

On the other hand, if such cut-offs are at both ends of the spectrum (20-30 and 18K-20K), wouldn't the ringing become hard to hear ?

The "ringing" is at the cut-off frequency. If this frequency is outside the audible range, then the "ringing" is inaudible and of no consequence. I've written a bit about this topic here: https://troll-audio.com/articles/filter-ringing/

 

[Jose Hidalgo]

If the ringing is at the cut-off frequency, and if that cut-off frequency is at one of the audible range ends, then it will indeed be hard to hear, won't it ? I guess ringing is already hard to hear on its own by untrained ears (you show it on your diagrams, its amplitude is much lower than the "pure" signal), but at such low or high frequencies I doubt we could hear anything.

Besides, I've read your article, and your conclusion is this :

ringing effects from oversampling filters can be eliminated by leaving a small margin above the highest frequency in a recording. Depending on the sample rate used, the filters needed to create this headroom may of course result in some ringing of their own. However, this is fully under the control of the producer. Even a very steep linear phase filter with long ripple tails will not add to whatever ringing, if any, is present in the distributed file.

Aren't you actually making Tks's point here ? He wanted to do just that : leave a small margin above the highest frequency (and the lowest too), by cutting off anything higher than, say, 18K. So if that cut-off can be achieved, wouldn't it be logical to say that ringing, if any, should be negligible ?

The question on how to create such steep filters with a parametric EQ such as EAPO remains though...

 

[Tks]

If the ringing is at the cut-off frequency, and if that cut-off frequency is at one of the audible range ends, then it will indeed be hard to hear, won't it ? I guess ringing is already hard to hear on its own by untrained ears (you show it on your diagrams, its amplitude is much lower than the "pure" signal), but at such low or high frequencies I doubt we could hear anything.

Besides, I've read your article, and your conclusion is this :

Aren't you actually making Tks's point here ? He wanted to do just that : leave a small margin above the highest frequency (and the lowest too), by cutting off anything higher than, say, 18K. So if that cut-off can be achieved, wouldn't it be logical to say that ringing, if any, should be negligible ?

The question on how to create such steep filters with a parametric EQ such as EAPO remains though...

I don't understand this concept of ringing tbh. If the ringing is going to be audibly present amongst the entire bandpass of frequencies of the recordings, then I have no answer for this problem. If the ringing starts it's harshest at the frequencies being decimated by the particularly strong high-pass and low-pass frequencies I chose for the 96dB/octave of sloping I gave as an example, then I do have an answer for this even if the ringing is quite harsh there.

If this ringing starts harshly at the chosen two frequencies and slowly lessens as we move from those two frequencies, the answer would be to just throw in a second PEQ pass sequentially after the first two (but with far less harsher sloping, so something like 46db/octave) to bury the ringing at those edges of the initial frequency.

So if we're going to do what I did before: 30Hz and 18kHz, and have both of those slope with a harsh 96dB/octave of high pass and low pass respectively and experience this 'ringing' ordeal. We THEN go for a second pass of two more frequencies (since the title of the thread allows for 4 frequencies) 35Hz and 17.5kHz sloping with either 48db/octave, or even lesser 24dB/octave to push out any of the ringing that would be occuring at the initial harsh-slopped 30Hz and 18kHz frequencies.

But I'm unaware how audible any of this ringing stuff is, and if this is present within audible concern of both low pass AND high pass filtering of frequencies, or is it just people claiming this is a problem with low pass filters due to high frequencies mainly being responsible for this undesirable audible ringing?

 

[DavidMcRoy]

Do you know if any affordable PC based program allows tilt manipulation? I don't think EQ APO or peace gui have this.

I found this function in sonar works but price is steep. This page has a cool animation for tilt control: https://www.sonarworks.com/soundid-...most-out-of-sonarworks-reference-4s-settings/

You can use a multi band PEQ to do it (if that’s all you have at your disposal) by just progressively altering levels per band up or down, centered on around 1.5 kHz or so.

 

[DavidMcRoy]

For a look at the old Quad preamp’s tilt control, see this link:

https://www.kenrockwell.com/audio/quad/34-preamplifier.htm

When you play with ”tilt,” you’ll discover that (a) a little goes a long way, and (b) you’re very likely to find a setting you subjectively “prefer” on a given source rather quickly. It’s intuitive, kind of like focusing a lens on a camera.

 

[zermak]

Aren't you actually making Tks's point here ? He wanted to do just that : leave a small margin above the highest frequency (and the lowest too), by cutting off anything higher than, say, 18K. So if that cut-off can be achieved, wouldn't it be logical to say that ringing, if any, should be negligible ?

The question on how to create such steep filters with a parametric EQ such as EAPO remains though...

Here is a start. 4x of it in EqualizerAPO makes a 96dB/oct filter. I am using it myself to cut off all crap over 20kHz, like @Tks mentioned indeed.

 

[Jose Hidalgo]

Thanks. Funny thing, I was talking to Juha myself yesterday before reading your post. He was very helpful, so I've got that part already figured out now

So, good news : I now have the technical tools to perform all the potential enhancements discussed in this thread : very steep filters, linear tilt, bass boost, warmth, clarity, etc.. Now I will go back to development, and hopefully everything will be ready shortly for you to test and enjoy. Stay tuned !

Oh, BTW...

Airiness is an upper treble thing

You are absolutely right. The word I was looking for was "Clarity", but I mispelled it.

 

[Jose Hidalgo]

Hi all. I come back to report as promised. I have successfully added a custom EQ to my upcoming EAPO-based app, to further adjust sound after you've matched one of the preset target curves specific to your device (especially for headphones, thanks to AutoEQ, Oratory1990 and such).

It works perfectly so far. Here's the current slider-based GUI. It started as 5 sliders (3 + Boost + Tilt), then 6, then 7 for visual symmetry around the "1K Tilt" :

Now on to the various filters that are open to discussion :

• Boost : is a 2-filter pack that is close to a 100 Hz low-shelf, but slightly improved. It reacts differently depending on the plugged device (it's not the same for AE-OE headphones, IE headphones or loudspeakers )
• Warmth : is a 3-filter pack in the 190-400 Hz range for vocal frequencies, both male and female.
• Box : is a 600 Hz peak that can help achieve balance between "boxy" and "hollow".
• 1K Tilt : as requested, this tilts the whole audible range in a linear way, as follows : 20 Hz @ X dB -> 1 KHz @ 0 dB -> 20 KHz @ -X dB. X can be positive or negative.
• Clarity : is a 2250 Hz peak. It was a 3-filter pack before, maybe wrongly placed at 3-4 KHz.
• Finesse : is a 5000 Hz peak.
• Airiness : is a 10 KHz high-shelf.

Your comments in this thread helped a lot, plus I relied on some documentation such as this chart :

Once again, this is not a regular EQ, but an EQ that's supposed to help you easily modify the sound after it has matched a target curve. Its effect is generally subtle but noticeable. It's supposed to help on a song-per-song basis, and my personal testing confirms that it does indeed ! The difference is really welcome on a number of songs.

Just for peace of mind, I'll be happy to read your comments and see if the frequencies, Q-factors, etc., can be further improved before I release the app (or after - beta-testers are still welcome BTW, a new round of testing is about to begin). Thanks in advance !

 

[DavidMcRoy]

Now that you are proposing 7 bands instead of just 4, I have to reassess my suggestions.

I still feel strongly about the need for a cut at 3500Hz to reduce coldness and push the soundstage back if desired. I’d replace “Clarity” with this if necessary, as a boost at 3.5kHz could alternately improve vocal articulation if needed.

A 3dB boost at 8000Hz is helpful to make highs sound cleaner and more liquid. I would replace your 5000Hz peak with this, since a boost at 5kHz will just make vocals edgier and harsh. if I had to I’d sacrifice “Finesse” for a boost at 8kHz.

 

[Jose Hidalgo]

Yes, actually I just want to avoid having two sliders too close in frequency. I would gladly trust people with more experience than me on the EQ side.

I went for 2250 and 5000, but it you feel that 3500 and 8000 would be more useful I can easily change these two sliders (that would just mean that there would be nothing between 600 and 3500). I just wonder how I should call them. I need short names for them that mean something to people (like "Clarity" and "Finesse"), I don't want to just write down frequencies on the GUI.

Once again for everybody, this EQ is intended to be performed after matching a given target (e.g. Harman), on a track-by-track basis.

 

[AryaStealth]

I noticed that Jose Hidalgo found a solution for Low and High Pass filters for EqualizerAPO. I'm not sure if these filtres go as high as 100dB+/octave:
Here how it looks like:

# Extremes Cutoff for EqualizerAPO

Filter 1: ON HPQ Fc 30 Hz Q 0.50241929
Filter 2: ON HPQ Fc 30 Hz Q 0.52249861
Filter 3: ON HPQ Fc 30 Hz Q 0.56694403
Filter 4: ON HPQ Fc 30 Hz Q 0.64682178
Filter 5: ON HPQ Fc 30 Hz Q 0.78815462
Filter 6: ON HPQ Fc 30 Hz Q 1.0606777
Filter 7: ON HPQ Fc 30 Hz Q 1.7224471
Filter 8: ON HPQ Fc 30 Hz Q 5.1011486

Filter 9: ON LPQ Fc 18000 Hz Q 0.50241929
Filter 10: ON LPQ Fc 18000 Hz Q 0.52249861
Filter 11: ON LPQ Fc 18000 Hz Q 0.56694403
Filter 12: ON LPQ Fc 18000 Hz Q 0.64682178
Filter 13: ON LPQ Fc 18000 Hz Q 0.78815462
Filter 14: ON LPQ Fc 18000 Hz Q 1.0606777
Filter 15: ON LPQ Fc 18000 Hz Q 1.7224471
Filter 16: ON LPQ Fc 18000 Hz Q 5.1011486