Sampling rate vs output question

[Grant Slack]

8 hours ago, MLXXX said:

A commonly sought after goal for hi-fi sound reproduction is to recreate the sound of a live performance.

 

If healthy human ears add small amounts of harmonic and intermodulation distortion when listening to a live performance, we accept that as "natural".

 

If something in the recording or playback process results in an increase in the level of harmonic or intermodulation distortion beyond what a listener would experience at a live performance, that would be an unnatural reproduction, and it might well be classified as undesirable from an academic viewpoint.   On the other hand, some people may actually prefer additional HD and/or IMD, such as from vinyl discs or loudspeakers systems, and may describe these distortions as "warmth".

 

It has been conventional wisdom that because human hearing cannot detect a test tone that exceeds 18kHz or so, that it is unnecessary (a waste of resources)  to capture and reproduce audio frequencies above 20kHz.

 

What if an experimental result suggests that a harmonic in the inaudible range makes a subjective difference? That was the outcome of the study published in 2008 entitled  "Temporal resolution of hearing probed by bandwidth restriction", previously mentioned in this thread. In that study, a 7kHz squarewave was listened to with earphones (capable of an extended response into the ultrasonic) either with its harmonics attenuated with a first order resistor-capacitor network, or not attenuated.  With attenuation the sound was reportedly dull. The fundamental was attenuated to a very minor extent but the third harmonic (21kHz) was attenuated to a considerable extent and phase shifted. [The 7kHz waveform being a squarewave, its harmonics were all odd either with or without the low pass RC filtering being applied.]

 

Has that paper already been discussed on this forum? I cannot find the name of the paper in a search, but I'm not sure all SNA content is searchable.

 

The paper itself advanced a number of possible explanations including that non-linearities in the ear produced an in phase even harmonic (14kHz) from the 7kHz fundamental, and that interacted with the delayed phase 21kHz harmonic signal reproduced by the earphones.

 

________

 

It wouldn't invalidate an apparent need to reproduce an ordinarily inaudible harmonic if the reason human ears could detect the harmonic was distortion arising in the ear, provided that distortion was normal for a healthy human ear.

 

It wouldn't support a need to reproduce an ordinarily inaudible harmonic if the  addition of the test harmonic created a discernible frequency, in the air in the listening room, that was an ordinarily audible frequency.

 

Example of an invalid testing method:

A loudspeaker was  used to reproduce an 8kHz tone together with a sporadic 24kHz tone. The 24kHz tone when present created a 16kHz tone at a non-negligible level (through loudspeaker IMD).

Hello MLXXX,

 

from your description, I am far from satisfied that the effect of the first-order filter on the audible bandwidth was not audible. In which case, the dull sound could be entirely explained by conventional means. At which point, Occam's Razor comes into play, and the "number of possible explanations" becomes moot.

 

cheers

Grant

[MLXXX]

26 minutes ago, Grant Slack said:

from your description, I am far from satisfied that the effect of the first-order filter on the audible bandwidth was not audible.

There was a very slight reduction in the amplitude of the fundamental, so slight as to be very unlikely to have made an audible difference of itself.

 

As for how the earphones could have produced distortion products in the audible range, rather than simply reproducing the 3rd and 5th harmonics of 7kHz, in addition to the fundamental, is not clear. The study actually took measurements of the earphone output and no extraneous products of significance were identified. Let me try and find a relevant part of the paper...

 

Ok, this paragraph from section C of Temporal resolution of hearing probed by bandwidth restriction seems relevant:

 

The presence of ultrasonic components at high transducer driving levels and inadequate damping can produce anharmonic distortion products within the audible band (Ashihara and Kiryu, 2000). To rule out the role of anharmonic distortion, subharmonics, noise, and other spurious components in the audible band, the spectrum of the signal was also measured separately using unaveraged signals (since synchronized averaging attenuates anharmonic frequencies). This power spectrum is shown in Fig. 4. No subharmonic peaks could be distinguished from noise; the absolute sound level of this noise in the 3.5 kHz subharmonic vicinity is <0 dB SPL (its fractional change due to filtering at threshold would be 0.046 dB). The collective power in all sub-fundamental frequencies in the 20 Hz–6 kHz band is less than 0.04% of the power in just the fundamental peak (7 kHz ± 0.1 kHz band). This freedom from anharmonic distortion is ensured by the superior quality of the electronics and transducer, and the relatively moderate (69 dB SPL) level of operation. Also because the entire signal chain is analog, spurious frequencies that can result from aliasing in digital systems are avoided. As will be shown below, the linearity—as reflected by the attenuations in the harmonic coefficients—is essentially perfect. 

Edited September 18, 2019 by MLXXX

[Audiophile Neuroscience]

7 hours ago, LHC said:

 

 

@Audiophile Neuroscience the full PDF download from here (https://www.researchgate.net/publication/313228333_High-Resolution_Audio_with_Inaudible_High-Frequency_Components_Induces_a_Relaxed_Attentional_State_without_Conscious_Awareness) is actually a full proceeding of papers dealing with the subject of music and the brain. You may have read some of these papers already. 

Thanks LHC, I certainly have some of these references and at least a couple dozen on Oohashi's "Hypersonic effect". Most studies seem to point to both behavioral responses and brain activity responses such as EEG alpha waves or other correlates indicating a perception has occurred in relation to ultrasonic otherwise inaudible sounds.

 

Interestingly there appears to be a "lag phase" for some of the effects which would seem to correspond to many people's experience for preferring longer time periods to evaluate sound quality, irrespective of the mechanism underpinning the reported differences/preferences. Also, the notion that these perceptions may take place below the level of overt awareness but there is a cumulative effect over time appears to be consistent with the above. IMO this should not be interpreted as being the same as therefore a negligible difference (although some will).

 

A bit of a digression and a stretch regarding relevance is the so called Iowa gambling task where subjects are able to sense a difference in a bad deck of cards with as yet inadequate information to conclude same. Nonetheless they somehow sense it and display autonomic nervous system changes that correlate with the uneasy feeling evoked by the task. This is more about decision making and responses in frontal lobe and amygdala etc but is interesting in the context of how we respond to incoming information that has not quite been fully consciously defined. Call it "gut feel/sensation" if you like

 

Whatever the mechanism, and not suggesting "gut feel" literally or otherwise or for that matter bone conduction over air conduction etc, the other interesting aspect (to me) is that some speculate that the mechanism of hypersonic modulation relates somehow to the initial attack of transients in the musical envelope and associated high frequency component and rise time of the corresponding frequencies extending above 20 Khz (Fourier analysis of a sharp transient attack envelope involving rapid onset of sound and short time to peak amplitude involve high frequency waveforms with rapid signal rise times.)

 

That theory marries nicely with what we know about attack transients being important in influencing our perception of tone/timbre which in turn relates to harmonic content and the dynamics of the attack-delay characteristics of the sound.

 

 

 

 

Edited September 18, 2019 by Audiophile Neuroscience

[Grant Slack]

6 hours ago, Muon N' said:

I doubt most systems ability to reproduce all of the harmonic structure of instruments, how many can do so? Can any truly?

My view is a resounding...no.

 

Hello Muon,

 

to quote the dead fellow's hologram in I Robot the movie, "You must ask the right questions."

 

IMO the right question is, "does the hifi need to reproduce the harmonic structure of instruments that is not audible?" 

 

My view is a resounding...no.

 

And my view is shared by the best audio engineers and researchers, AFAICT.

 

@MLXXX is exploring the possibility that some experiments might hint otherwise, with special test tones. This exploration is very, very far away from constituting an argument that reproducing the inaudible harmonic structure of instruments will bring any benefit to home music listening. IIRC MLXXX has said as much, several times.

 

Regards,

Grant

[Guest Muon N']

I disagree heartily!

 

Also, I was also including all harmonics of that between 20hz and 20kHz.

[Grant Slack]

15 minutes ago, Audiophile Neuroscience said:

Thanks LHC, I certainly have some of these references and at least a couple dozen on Oohashi's "Hypersonic effect". Most studies seem to point to both behavioral responses and brain activity responses such as EEG alpha waves or other correlates indicating a perception has occurred in relation to ultrasonic otherwise inaudible sounds.

 

Interestingly there appears to be a "lag phase" for some of the effects which would seem to correspond to many people's experience for preferring longer time periods to evaluate sound quality, irrespective of the mechanism underpinning the reported differences/preferences. Also, the notion that these perceptions may take place below the level of overt awareness but there is a cumulative effect over time appears to be consistent with the above. IMO this should not be interpreted as being the same as therefore a negligible difference (although some will).

 

....Whatever the mechanism, and not suggesting "gut feel" literally or otherwise or for that matter bone conduction over air conduction etc, the other interesting aspect (to me) is that some speculate that the mechanism of hypersonic modulation relates somehow to the initial attack of transients in the musical envelope and associated high frequency component and rise time of the corresponding frequencies extending above 20 Khz (Fourier analysis of a sharp transient attack envelope involving rapid onset of sound and short time to peak amplitude involve high frequency waveforms with rapid signal rise times.)

 

That theory marries nicely with what we know about attack transients being important in influencing our perception of tone/timbre which in turn relates to harmonic content and the dynamics of the attack-delay characteristics of the sound.

 

 

 

 

Hello David,

 

I urge caution regarding Oohashi's "Hypersonic effect".

 

There has been more than one attempt to validate its existence, always unsuccessful, including by the technical research division of NHK, the Japanese national broadcaster.

 

I believe the most reasonable position to take wrt its existence, is highly dubious.

 

Hence, any theories that build on it have the same dubious status.

 

Regards,

Grant

[Grant Slack]

Hello Muon,

 

11 minutes ago, Muon N' said:

I disagree heartily!

 

A rational basis for disagreement would be greatly appreciated!

 

13 minutes ago, Muon N' said:

Also, I was also including all harmonics of that between 20hz and 20kHz.

I am comfortable that those harmonics are well reproduced by hifi gear.

 

Regards,

Grant

[Grant Slack]

28 minutes ago, MLXXX said:

There was a very slight reduction in the amplitude of the fundamental, so slight as to be very unlikely to have made an audible difference of itself.

 

Hello MLXXX,

 

nevertheless, the apparent lack of testing of that assumption by the researchers, would be a critical flaw.

 

cheers

Grant

[Guest Muon N']

I'm best not bothering with discussions that are at the end of day... of no use to myself.

Edited September 18, 2019 by Muon N'
typo

[MLXXX]

3 hours ago, Grant Slack said:

Hello MLXXX,

 

nevertheless, the apparent lack of testing of that assumption by the researchers, would be a critical flaw.

 

cheers

Grant

I note that even if it could be established that a 0.18dB reduction in sound level of a 7kHz tone was audible, that wouldn't seem to explain the subjective report of a duller sound.

 

One thing that did concern me was that it was reported that subjects needed several seconds' exposure to a tone. Section D of the paper includes this:

It was found in preliminary testing (especially when τ is close to the threshold) that subjects needed to listen to the tones for several seconds to form a lasting impression of the sounds; immediately after switching the subjects had difficulty assessing whether anything had changed or not. This again confirms that the gating itself does not provide a cue.

 

It occurred to me that there could be some sort of indirect fatigue effect from the presence of the 21kHz harmonic. I'd have been more impressed if the difference in the sharpness or dullness had been perceived immediately, rather than requiring several seconds to become apparent.

Edited September 18, 2019 by MLXXX

[Audiophile Neuroscience]

49 minutes ago, Grant Slack said:

Hello David,

 

I urge caution regarding Oohashi's "Hypersonic effect".

 

There has been more than one attempt to validate its existence, always unsuccessful, including by the technical research division of NHK, the Japanese national broadcaster.

 

I believe the most reasonable position to take wrt its existence, is highly dubious.

 

Hence, any theories that build on it have the same dubious status.

 

Regards,

Grant

Let's just say it is controversial.

 

I am guessing the NHK paper you cite mentioned was the 2003 Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components by Nishiguchi et al and which commented on the 2000 Oohashi paper. I agree with the conclusion "The results showed that we can still neither confirm nor deny the possibility that some subjects could discriminate between musical sounds with and without very high frequency components."

 

Nishiguchi went on to publish other papers such as Perceptual discrimination of very high frequency components in wide frequency range musical sound Author(s): Nishiguchi et al Source: APPLIED ACOUSTICS  Volume: 70   Issue: 7   Pages: 921-934 JUL 2009

"Tests showed that the subjects could discriminate between musical sounds with and without very high frequency components [above 20KHz] . This paper describes these subjective evaluations in terms of reproducing such very high frequency components in musical Sound."

 

Neither of those NHK papers AFAIK performed neurological testing to correlate with the behavioral responses.

 

There have been numerous studies since that do indicate neurological and behavioral responses.

Edited September 18, 2019 by Audiophile Neuroscience

[Grant Slack]

...all straight out of Wikipedia  

 

cheers

Grant

[Guest Muon N']

I think I'm siding with David here, he has a much better handle on this.

 

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0095464

[Grant Slack]

...and again that's Oohashi, so standard caveats apply.

 

Your linked 2014 report is even more interesting, though. Its claim is that the Hypersonic 'mood effect' is negative up to 32 kHz, then positive above 32 kHz.

 

So, considering that ultrasonic sound energy from musical instruments is rapidly diminishing with frequency, then what little there is above 20 kHz will mostly lie below 32 kHz, resulting in the mood effect being far more likely to be negative than positive!

 

Which is the opposite of Oohashi's original work. He has moved on, it seems.

 

It would also be nice to see some research into this using more common music than Oohashi's insistence on gamelan-intensive music, with its very uncommon spectrum with a lot of ultrasonic energy. May I initially suggest that he would not get a result with typical musical instruments.

 

But to repeat: this 2014 report would imply that we will generally have a more positive mood response if we filter out signal above 20 kHz. Hardly an endorsement of high-res distribution! 

 

cheers

Grant

[Guest Muon N']

*shakes head*

[LHC]

6 hours ago, Audiophile Neuroscience said:

the other interesting aspect (to me) is that some speculate that the mechanism of hypersonic modulation relates somehow to the initial attack of transients in the musical envelope and associated high frequency component and rise time of the corresponding frequencies extending above 20 Khz (Fourier analysis of a sharp transient attack envelope involving rapid onset of sound and short time to peak amplitude involve high frequency waveforms with rapid signal rise times.)

 

That theory marries nicely with what we know about attack transients being important in influencing our perception of tone/timbre which in turn relates to harmonic content and the dynamics of the attack-delay characteristics of the sound.

 

And that is why recent debate has shifted away from trying to 'hear' hypersonic frequencies in the frequency domain, to arguments over time (or temporal) resolution in the time domain (and this is different to what some would called 'timing' resolution). Maybe by happenstance, the time smearing debate in the time domain also coincided with an emerging interests in the reconstruction filters that can harm temporal resolution. It did not help to have MQA drop on the scene and their marketing created more confusion and resentment. Anyway I was glad those 2019 AES papers I posted early in the thread made an attempt to clear the air and lay down a clear picture of where we are, and where they want to go with hi res.

 

Anyway back to your post. Your intuition matches nicely the reasoning given in the new 2017 paper (pp127-128):

 

"Second, the underlying mechanism of how inaudible high-frequency components affect EEG activities cannot be revealed by the current data. It is noteworthy that presenting high- frequency components above 20 kHz alone did not produce any change in EEG activities (Oohashi et al., 2000). Therefore, the combination of inaudible high-frequency components and audible low-frequency components should be a key factor that causes this phenomenon. A possible clue was obtained by a recent study of Kuribayashi and Nittono (2015). Recording sound spectra of various musical instruments, they found that high- frequency components above 20 kHz appear abundantly during the rising period of a sound wave (i.e., from the silence to the maximal intensity, usually less than 0.1 s), but occur much less after that. Artificially cutting off the high-frequency components may cause a subtle distortion in this short period. It will take some time to accumulate these small, short-lasting differences until they produce discernible psychophysiological effects. This explanation is consistent with the fact that the effect of high- frequency components on EEG activities appears only after a 100–200-s exposure to the music (Oohashi et al., 2000, 2006; Yagi et al., 2003a; Fukushima et al., 2014; Kuribayashi et al., 2014; Ito et al., 2016)."

 

I underlined a key insight here: it is not about those inaudible frequencies alone, but rather the combined effect of the whole range of frequency components in the time domain that really matters. 

[Audiophile Neuroscience]

28 minutes ago, LHC said:

(Oohashi et al., 2000, 2006; Yagi et al., 2003a; Fukushima et al., 2014; Kuribayashi et al., 2014; Ito et al., 2016)."

 

I underlined a key insight here: it is not about those inaudible frequencies alone, but rather the combined effect of the whole range of frequency components in the time domain that really matters. 

I agree with you....or OTOH we could just go with Wikipedia

[LHC]

2 hours ago, Audiophile Neuroscience said:

I agree with you....or OTOH we could just go with Wikipedia

Maybe I am a bit slow, but I don't get the joke regarding the wikipedia reference? Is this the relevant page https://en.wikipedia.org/wiki/Hypersonic_effect?

 

I note from that wiki page the following point: "Studies cited as contrary evidence did not address the physiological brain response to high-frequency audio, only the subject's conscious response to it." The new 2017 study adds to the 'physiological brain response' evidence, and acknowledges its test subjects cannot consciously hear any difference. So it is consistent with wikipedia. 

Edited September 18, 2019 by LHC

[Audiophile Neuroscience]

Hi LHC, just a little fun with Grant's cited evidence...

hours ago, Grant Slack said:

There has been more than one attempt to validate its existence, always unsuccessful, including by the technical research division of NHK, the Japanese national broadcaster.

 

21 hours ago, Audiophile Neuroscience said:

I am guessing the NHK paper you cite mentioned was the 2003 Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components by Nishiguchi et al

18 hours ago, Grant Slack said:

...all straight out of Wikipedia  

[MLXXX]

Just a quick overview comment, in case some people skimming through this thread can't understand why people are taking such an intense interest in sample rate.

 

It really is a fundamental question for an audiophile to establish whether 44.1kHz is a sufficient sample rate for music. If it isn't then we really should seriously consider accessing only "hi-res" versions of music wherever possible.

 

For over a century it's been known that adult human beings can't hear over 18kHz or so for continuous tones presented in isolation.

 

Relatively recent work has focused on whether ultrasonic frequencies might play a role in aiding perception of very short delays as between the soundwaves presented to one ear and the other; and whether changes in the amplitude or phase of a harmonic inaudible if presented in isolation might be perceptible when presented to the ears at the same time as the fundamental.

 

And there has been work on whether ultrasonics despite not being audible can somehow alter a person's mood.

 

I think it's fair to say that these more recent studies have not led to persuasive evidence that 44.1kHz is deficient as a  sample rate for music (as distinct from audio consisting of specially selected test tones).

 

That is reassuring for those using CDs to listen to music.  It is of course not helpful for an industry wishing to market hi-res audio as a necessary, and clearly superior, distribution format.

[Audiophile Neuroscience]

12 hours ago, LHC said:

"Studies cited as contrary evidence did not address the physiological brain response to high-frequency audio, only the subject's conscious response to it." The new 2017 study adds to the 'physiological brain response' evidence,

Yes, I agree, this and other studies combining a behavioral and neurological response suggest perception has occurred. This had previously been demonstrated for "resting" or inactive subjects.

 

As noted in the 2017 study increased alpha activity is seen during a relaxed attentional state and thought to be a signifier of enhanced processing, and Beta activity has been shown to increase when arousal and vigilance level increase.

 

Kuribayashi and Nittono (2014)  found that "the difference between full-range and high-cut sounds appeared in the right inferior temporal cortex,  whereas  the main source of the alpha-band activity was located in the parietal-occipital region."

 

"The finding that the alpha-band activity difference was obtained in specific but not whole regions is suggestive that this increase may reflect an activity related to task performance rather than a global arousal effect (Barry et al.2007)."

 

In the 2017 study they felt that EEG and Event related potentials supported  high-resolution audio with inaudible high-frequency components "enhances the cortical activity related to the attention allocated to task-relevant stimuli."