• From: Germany
  • Registered: 2022-02-13
  • Posts: 69

Topic: Interpolation Filter audible?

On page 86 of the ADI-PRO da R manual you will find the different Impulse behaviors of the filters. Does this behavior apply over the entire frequency spectrum or only for the upper frequencies near the filter frequency?

I have also discussed this topic with people on ASR because I claim that I hear a difference between Fast Linear and Fast Min Phase filters, especially at low frequencies. (Bass and kickdrum)

They think I'm just imagining it and that the pre and post ringing is not audible because there is not much signal near the Nyquist frequency to cause ringing.
That confuses me a lot.

I'm not the only one who hears differences, even my brother who has no idea about audio and also people here in the RME forum hear the difference...

  • Registered: 2023-06-30
  • Posts: 161

Re: Interpolation Filter audible?

You seem to confuse the filter's frequency response (frequency domain) with its impulse response (time domain). Impulse is a short and broadband signal.

Fireface UCX II + ARC USB > ADI-2 Pro FS R BE > Neumann KH 750 DSP + MA 1 > KH 120 A
  • From: Germany
  • Registered: 2022-02-13
  • Posts: 69

Re: Interpolation Filter audible?

unpluggged wrote:

You seem to confuse the filter's frequency response (frequency domain) with its impulse response (time domain). Impulse is a short and broadband signal.

I know, my question is whether the changed impulse behavior in the illustration of the ADI-2 manual applies to the entire frequency spectrum or does it change depending on the frequency?

  • rawac
  • AES/EBU artist
  • Offline
  • From: Germany
  • Registered: 2020-09-23
  • Posts: 122

Re: Interpolation Filter audible?

A perfect pulse contains all frequencies, theoretically. In practice, no system can generate all frequencies, which is why real pulses only contain frequencies up to a certain limit. The same applies to the processing of pulses.

Theoretically, every filter always affects all frequencies. In practice, however, the effect on certain frequency ranges is so small that it can easily be neglected.  This also applies to the setting of the DA Impulse Responses, which are also filters. The only difference is that with these filters, the time behaviour is the main focus of the design and not the frequency response.

The effect that the different settings have on the frequency response is shown in the manual, see chapter 34.14 DA Filter curves 44.1 kHz. Therefore, it is not clear to me what the question is here, or whether it is really just a lack of basic knowledge.

Ralf
(ADI-2 Pro FS with ThinkPad Yoga L13, Dynaudio Focus 600 XD or Focal Clear — and a lot of Jazz)

rawac's Website

5 Reply by KaiS (edited by KaiS 2024-01-13 17:06:31)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,525

Re: Interpolation Filter audible?

hasan.ay386 wrote:
unpluggged wrote:

You seem to confuse the filter's frequency response (frequency domain) with its impulse response (time domain). Impulse is a short and broadband signal.

I know, my question is whether the changed impulse behavior in the illustration of the ADI-2 manual applies to the entire frequency spectrum or does it change depending on the frequency?

The clear answer is:
Low frequency pulses pass practically unchanged, no matter what filter is used.

The opposite is true for high frequencies, all filters are different, and do sound different - but only audible for audio at sample rates of 44.1 and 48 kHz.


The filter’s differences are:

• Frequency response: “Slower” filters have less treble energy on sustained signals, but higher output on pulses.

• Impulse response: “Slower” filters have less ringing.

• Shape of impulse response: “minimum phase” filters have less unnatural pre-ringing, opposed to “linear phase” filters, which have a high amount of unnatural pre-ringing.
(Remark: every musical instrument has post-ringing, resonance, but no instrument sounds before it’s played!).

• Aliasing artifacts: Too slow filters, or almost no filtering like the “NOS-Filter” causes too much aliasing artifacts that reach and smear the whole audio band.
Aliasing is a very strange, unnatural sound, the opposite of “ black background”.


So, selection of a DA-Filter is a compromise, kind of.

Subjectively changes in the treble seem to affect the bass, but that’s only relative.
Bass in music doesn’t ever appear isolated, and a bass-drum e.g. has significant energy even in the treble, so …


Personally my favorites filters are “Slow” on AKM-DA-chips and SD LD (Low Dispersion) on ESS-DA-chips.

Same named filters on AKM and ESS are not the same.

KaiS's Website

  • Kubrak
  • DAW Mastermind
  • Offline
  • Registered: 2023-08-03
  • Posts: 548

Re: Interpolation Filter audible?

The kick may also have high frequencies, but just for a short time.

7 Reply by KaiS (edited by KaiS 2024-01-13 22:44:25)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,525

Re: Interpolation Filter audible?

Kubrak wrote:

The kick may also have high frequencies, but just for a short time.

Kick = Bassdrum.

There‘s no natural instrument without overtones / treble.
The overtones and the attack phase define the character of the instrument, much more then a sustained second part of a musical note.

Every musical sound has a pulsive character by some amount (it does “start” somewhere), oppressed to a constant sine wave for measurement.
Every pulse, by physics, contains higher frequencies that are subjected to the influence of the DA-filter.

KaiS's Website

8 Reply by hasan.ay386 (edited by hasan.ay386 2024-01-14 01:55:54)

  • From: Germany
  • Registered: 2022-02-13
  • Posts: 69

Re: Interpolation Filter audible?

KaiS wrote:

The clear answer is:
Low frequency pulses pass practically unchanged, no matter what filter is used.

The opposite is true for high frequencies, all filters are different, and do sound different - but only audible for audio at sample rates of 44.1 and 48 kHz.


The filter’s differences are:

• Frequency response: “Slower” filters have less treble energy on sustained signals, but higher output on pulses.

• Impulse response: “Slower” filters have less ringing.

• Shape of impulse response: “minimum phase” filters have less unnatural pre-ringing, opposed to “linear phase” filters, which have a high amount of unnatural pre-ringing.
(Remark: every musical instrument has post-ringing, resonance, but no instrument sounds before it’s played!).

• Aliasing artifacts: Too slow filters, or almost no filtering like the “NOS-Filter” causes too much aliasing artifacts that reach and smear the whole audio band.
Aliasing is a very strange, unnatural sound, the opposite of “ black background”.


So, selection of a DA-Filter is a compromise, kind of.

Subjectively changes in the treble seem to affect the bass, but that’s only relative.
Bass in music doesn’t ever appear isolated, and a bass-drum e.g. has significant energy even in the treble, so …


Personally my favorites filters are “Slow” on AKM-DA-chips and SD LD (Low Dispersion) on ESS-DA-chips.

Same named filters on AKM and ESS are not the same.

Thanks! This answers my question very precisely.

9 Reply by MC

  • MC
  • MC
  • Administrator
  • Online
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,586

Re: Interpolation Filter audible?

I think the point is: the specific ringing shown only happens with frequencies near SR/2. These frequencies are either non-existent, or very low in amplitude, or extremely short in time, so ringing at 22 kHz is a non-issue. Especially as this ringing is 22 kHz itself, so it is inaudible as well.

The example of a bass drum really bothers me, as it is so typical for wrong wording and thinking. A bass drum might subjectively have an attack (hence the association with 'kick'), but look at its waveform (a slow swinger even at the start of the waveform), and check the spectral energy - it is dead above 5 kHz. There is no real 'attack' at all, and absolutely nothing that could excite a filter ringing at 22 kHz.

Regards
Matthias Carstens
RME

MC's Website

  • Kubrak
  • DAW Mastermind
  • Offline
  • Registered: 2023-08-03
  • Posts: 548

Re: Interpolation Filter audible?

But what if it is unnatural electronic bass drum? It may go sharply up  (within one, two samples) and so contain high frequencies. And if it is not mastered well....

But I do not speak from my experience, just recalling what math says....

11 Reply by hasan.ay386 (edited by hasan.ay386 2024-01-14 18:50:53)

  • From: Germany
  • Registered: 2022-02-13
  • Posts: 69

Re: Interpolation Filter audible?

MC wrote:

I think the point is: the specific ringing shown only happens with frequencies near SR/2. These frequencies are either non-existent, or very low in amplitude, or extremely short in time, so ringing at 22 kHz is a non-issue. Especially as this ringing is 22 kHz itself, so it is inaudible as well.

The example of a bass drum really bothers me, as it is so typical for wrong wording and thinking. A bass drum might subjectively have an attack (hence the association with 'kick'), but look at its waveform (a slow swinger even at the start of the waveform), and check the spectral energy - it is dead above 5 kHz. There is no real 'attack' at all, and absolutely nothing that could excite a filter ringing at 22 kHz.

So why do I hear a difference? (with headphones)

As an example in this sample, especially at the decay of the third note:
https://drive.google.com/file/d/17emPqb … share_link
Or this:
https://drive.google.com/file/d/1MMu_Rb … share_link

12 Reply by KaiS (edited by KaiS 2024-01-15 08:14:41)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,525

Re: Interpolation Filter audible?

MC wrote:

I think the point is: the specific ringing shown only happens with frequencies near SR/2. These frequencies are either non-existent, or very low in amplitude, or extremely short in time, so ringing at 22 kHz is a non-issue. Especially as this ringing is 22 kHz itself, so it is inaudible as well.

The example of a bass drum really bothers me, as it is so typical for wrong wording and thinking. A bass drum might subjectively have an attack (hence the association with 'kick'), but look at its waveform (a slow swinger even at the start of the waveform), and check the spectral energy - it is dead above 5 kHz. There is no real 'attack' at all, and absolutely nothing that could excite a filter ringing at 22 kHz.

The real world situation is more complex.

Bass drums or kick drums usually are layered with electronic drum sounds, and they almost never appear alone within the music, 99% there is a hihat or cymbal or percussion hit on top.


But that’s no my main point.
I can clearly distinguish between the various DA-Filters.

Trained by profession to listen analytical, I’d say the bass range is unaffected.
For me the main difference lies in definition and separation of higher pitched instruments and, connected to that, the ambience presentation.

The standard “Sharp” filters e.g. has some less differentiation between, e.g. hihats and shaker, even less space between their hits.
This effect even covers the ambience sounds.

I have to admit, the differences are subtle, and for me only appear at a sample rate of 44.1 kHz (I don’t have much 48 kHz music to listen to).
From 88.2 kHz and up I don’t hear any.

KaiS's Website