Are 48 kHz enough to non-bat humans?

[Miguelón]

Thinking about buying a pair of Neumann KH 120 2, and reading here and there opinions and specs I learned that their internal DAC samples at 48 kHz max.

As my DAC supports 192 kHz I suppose that the Neumann’s one will downsample to 48 kHz, but my question is more about signal fidelity.

As Nyquist theorem states, 48 kHz will be enough to reproduce all frequencies till 24 kHz, way more than my 48 years old ears can hear…

So can I expect any (audible, not theoretical) degradation of the source when passing my streamer to the Neumanns through the SPDIF output?

 

[Killingbeans]

So can I expect any degradation of the source when passing my streamer to the Neumanns through the SPDIF output?

Nope

 

[Miguelón]

Nope

Simple and direct, thanks!!

 

[DSJR]

As an aside, we in the UK used to rave about BBC live broadcasts on FM. Merely a 15kHz cut off on the 13 bit digital system they use(d).

 

[DVDdoug]

CDs at 44.1kHz are generally better than human hearing and if they aren't perfectly-flat to 20KHz they are usually "almost-flat" to 20kHz or beyond.

Presumably, the DSP (at 48kHz) is there to improve the sound (over what it would be without DSP) whether you're using the analog or digital input.

As Nyquist theorem states, 48 kHz will be enough to reproduce all frequencies till 24 kHz

It would be more precise to say it can't go over 24kHz. In the real world there are non-ideal anti-aliasing filters and for example you're not going to get 23,999Hz. And even without filters you get noticeable modulation as you get close the Nyquist limit. Well... noticeable if you can hear it... You can hear modulation if you generate a 3999Hz tone at a sample rate of 8kHz.

Of course it's "nice" if the equipment and the recording cover the "full traditional" 20-20kHz range.

I'm NOT arguing for lower performance, but the guys who researched psychoacoustics and designed lossy perceptual compression (MP3, etc.) discovered that even if you can hear to 20kHz in a hearing test, the highest harmonics in normal program material are masked (drowned-out) by slightly-lower frequencies. If you can hear that high, your hearing isn't as sensitive at the extremes, and any content at the highest frequencies is "weak" harmonics to begin with. If you hear compression artifacts in "high quality" MP3s, it's usually not the loss of high-frequencies that you hear.

 

[Robin L]

Understand that the 20khz limit is for young people. Once somebody is over 25 or so, high frequency perception deteriorates. I lost the top octave of my hearing at a Neil Young concert around 1978, when I was only 23. Could no longer hear the flyback transformers still in use with TVs at the time. They usually operated at 15 khz.

 

[Dunring]

I only have music in 44.1khz 16 bit flac so it plays well in other devices like the car stereo well. Unless you have a pet brown bear who wears headphones to listen to dog whistle quartets or nature documentaries, it covers all you can hear.

 

[Miguelón]

CDs at 44.1kHz are generally better than human hearing and if they aren't perfectly-flat to 20KHz they are usually "almost-flat" to 20kHz or beyond.

Presumably, the DSP (at 48kHz) is there to improve the sound (over what it would be without DSP) whether you're using the analog or digital input.


It would be more precise to say it can't go over 24kHz. In the real world there are non-ideal anti-aliasing filters and for example you're not going to get 23,999Hz. And even without filters you get noticeable modulation as you get close the Nyquist limit. Well... noticeable if you can hear it... You can hear modulation if you generate a 3999Hz tone at a sample rate of 8kHz.

Of course it's "nice" if the equipment and the recording cover the "full traditional" 20-20kHz range.

I'm NOT arguing for lower performance, but the guys who researched psychoacoustics and designed lossy perceptual compression (MP3, etc.) discovered that even if you can hear to 20kHz in a hearing test, the highest harmonics in normal program material are masked (drowned-out) by slightly-lower frequencies. If you can hear that high, your hearing isn't as sensitive at the extremes, and any content at the highest frequencies is "weak" harmonics to begin with. If you hear compression artifacts in "high quality" MP3s, it's usually not the loss of high-frequencies that you hear.

I’m quite sensible to compression algorithms, as I have a considerable visual impairment (not blind but enough to need hear to many things), even high quality MQA appears distorted to me.

I have the impression that has more to be with the phase space than the frequency, for example with headphones reproducing MP3, MQA and other loosy formats I tend to be confused in instrument location…

Not the same thing with monitors, but in general I use acoustic fields to improve my balance (for example I run with altered posture if I use headphones, and even walking with music my head and neck don’t move as normal).

Discussing this phenomenon with my ORL, we think that seem to be related with some capacity of perceive time delay within left and right side of reflected sound from the floor and walls.

Nevertheless the only way I have to test DSP impact on music scene is to hear it, in stereo of course because mono listenings don’t bothers when music is compressed.

Hope you’re right and even with analogue connections I won’t perceive artifacts.

If not, I will go with full analogue path with a pair of Genelecs 8030 that were my first choice, but Neumann specs are so high that who resist the temptation to try…

 

[Zapper]

I have the impression that has more to be with the phase space than the frequency, for example with headphones reproducing MP3, MQA and other loosy formats I tend to be confused in instrument location…

The KH120ii have excellent phase and temporal linearity. From the specs:

FIR phase correction	Linear phase (120 Hz ... 16 kHz; +/- 45°)

 

[Dialectic]

I formerly believed that high sampling rate audio (i.e., greater than 44.1 kHz) was desirable because 44.1/48 kHz.sampling could leave information missing between samples or somehow create misplaced samples.

When I had this strange belief, I didn't understand the Nyquist-Shannon sampling theorem, which the OP cites. All of the information you can hear can be represented by audio sampled at 48 kHz. The samples will not be in the wrong place, nor will anything that you can hear be missing.

I suspect the OP has a higher level of mathematical understanding than I did and will not repeat my ridiculous mistakes.

 

[ZolaIII]

It's not like bat but like cat! The noise shaping beyond hearing range is one thing and what you can hear is completely another. So in case of losy BT codecs it's a good thing and that's how LDAC high is almost lose less as usually they fall in uper highs and in hearing range. In case of lose less transmission anything over 20 KHz is not audible. The 20 KHz is if you have hearing of a new born and you don't have such. In reality up to 16 KHz is perfectly fine and there is little of it in recorded materials above 12~12.5 KHz anyway.

 

[MRC01]

48 kHz is enough. It is high enough that the transition band of the AA filter is wide enough, most DACs have near-perfect response. Unlike 44.1 kHZ, where the transition band is too narrow and most DACs either attenuate in the passband (below 20 kHz) or they stretch the transition band, not reaching full cut until around 24,100 which is above Nyquist and can leak HF noise.

So while 48 kHz is enough, my only concern would be if that device resamples everything to 48 kHz. Properly implemented high quality resamplers should be transparent. But some are not, especially those running with limited processing power doing non-integer rate conversions. I've encountered more than one resampler over the years that creates audible noise or artifacts. It's rare but it happens.

 

[jasonhanjk]

48kHz DAC can achieve 0.0005% thd+n (and lower), can the speaker internal amp achieve below 0.005%?
If no, then why bother? The amp creates artifact in your audible hearing range but you are concern about higher frequency?

 

[valerianf]

There is the simple law of evolution. At the start was 44.1 khz and 48 khz, then 96 khz and 192 khz.
Well, any equipment that is over capable is able to do less if needed.
Why to invest money in something that is limited?

Regarding picture quality, I try to watch 4k UHD DV movies, even if sometime I am watching 1080p or even 720p.
My Tv is 4k UHD DV capable as it should.
DAC in my computer or AVR are 192 khz capable.
It does not cost really more to get a hardware up-to-date.

 

[dfuller]

The only time you need to worry about high sample rates is if you are doing any processing that can introduce aliasing. Otherwise, don't worry about it.

 

[Miguelón]

I formerly believed that high sampling rate audio (i.e., greater than 44.1 kHz) was desirable because 44.1/48 kHz.sampling could leave information missing between samples or somehow create misplaced samples.

When I had this strange belief, I didn't understand the Nyquist-Shannon sampling theorem, which the OP cites. All of the information you can hear can be represented by audio sampled at 48 kHz. The samples will not be in the wrong place, nor will anything that you can hear be missing.

I suspect the OP has a higher level of mathematical understanding than I did and will not repeat my ridiculous mistakes.

Yes, I think the same!

When I play Apple Music or Tidal with the limitation of 44.1/ 48 kHz in the app settings, I can perceive a degradation in sound on tracks that are 44.1 kHz or 48 kHz native.

Conversely if I set streaming quality to the maximum they sound way more precise, even if in the app setting nothing tells that low sampling will affect even CD/ DVD quality tracks.

I mention Nyquist theorem because it states theoretical minimum to frequency sampling, but in general is misunderstood. Theorem specifies that to encode finite-composed function (finite in the sense that its Fourier transform is bounded) you will need AT LESAT double the highest frequency of its F. Transform to ensure lossless encoding.

In practice, as DVDdoug mentioned, the Nyquist frequency is a minimum requirement and in real world you need more than that. I’m not familiar with details of engineering applications of the theorem, but I hear what I hear: the highest sampling the better representation of the file regardless of the native sampling.

Neumann’s technician answered time ago that transducers at 96 kHz may affect drivers sound, as was mentioned in an article, so their monitors DAC are limited to 48 kHz.

 

[jasonhanjk]

There is the simple law of evolution. At the start was 44.1 khz and 48 khz, then 96 khz and 192 khz.
Well, any equipment that is over capable is able to do less if needed.
Why to invest money in something that is limited?

Regarding picture quality, I try to watch 4k UHD DV movies, even if sometime I am watching 1080p or even 720p.
My Tv is 4k UHD DV capable as it should.
DAC in my computer or AVR are 192 khz capable.
It does not cost really more to get a hardware up-to-date.

Not really a fair comparison.

A 4K TV vs 1080p TV, the difference is obvious and can be seen.

Whereas 48kHz vs 192kHz difference is so little. Both can produce 0.0003% and below THD+N. There's Chinese saying, remove pants before farting.

 

[Doodski]

Not really a fair comparison.

A 4K TV vs 1080p TV, the difference is obvious and can be seen.

Whereas 48kHz vs 192kHz difference is so little. Both can produce 0.0003% and below THD+N. There's Chinese saying, remove pants before farting.

There's also another sooper great Chinese saying that says that a nail that sticks up gets hammered down...LoL. It means a few things but I'll let you figure it out yourself.

 

[Miguelón]

48 kHz is enough. It is high enough that the transition band of the AA filter is wide enough, most DACs have near-perfect response. Unlike 44.1 kHZ, where the transition band is too narrow and most DACs either attenuate in the passband (below 20 kHz) or they stretch the transition band, not reaching full cut until around 24,100 which is above Nyquist and can leak HF noise.

So while 48 kHz is enough, my only concern would be if that device resamples everything to 48 kHz. Properly implemented high quality resamplers should be transparent. But some are not, especially those running with limited processing power doing non-integer rate conversions. I've encountered more than one resampler over the years that creates audible noise or artifacts. It's rare but it happens.

Yes, thanks for the information, is what I think also but had not enough experience to concrete my thoughts.

My streamer is the Wiim Pro Plus, I’m very satisfied with its quality but I have doubts about resampling on the Neumann’s DSP.

This give me the idea that perhaps is better to set my Wiim limited to 48 kHz to avoid unnecessary multiple resampling, is this correct?

 

[Miguelón]

48kHz DAC can achieve 0.0005% thd+n (and lower), can the speaker internal amp achieve below 0.005%?
If no, then why bother? The amp creates artifact in your audible hearing range but you are concern about higher frequency?

My concern is more about resampling, I agree with you that 48 kHz are enough