Very Low Freqs do affect or hearing (& perception)

I have long been of the opinion that noise is the great overlooked element in audio production - the focus tends to be on the signal itself & distortions, accuracy, etc. Less, maybe even very little consideration is given to sea of noise in which the signals are immersed & what affects this sea might have on the perceived sound. If it's not directly perceived then it's not often considered. But I've been of the opinion for a long time now that noise which isn't directly perceived can have a significant bearing on our auditory perception - CM noise & it's affects, for instance.

A recent paper published (Oct 2014) that begins to tie together some disparate strands of information about noise that I have held onto. The summary for this paper is : "Barely perceptible low-frequency signals nevertheless activate measurable responses in our auditory circuits. Neurobiologists have now characterized the remarkable impact of low-frequency sounds on the inner ear"

"The human auditory system appears to be poorly adapted to the perception of low-frequency sound waves, as hearing thresholds become markedly higher for frequencies lower than about 250 Hz. Yet sensory cells do react to pressure waves with frequencies below 100 Hz, as revealed by the fact that such signals actually evoke detectable micromechanical responses in nerve cells in the inner ear"

""It turns out that low-frequency sounds have a clearly definable modulatory influence on spontaneous otoacoustic emissions," says Drexl. Following exposure to the 30-Hz signal for 90 seconds, the subjects' SOAEs exhibited slow oscillations in frequency and level, which persisted for up to 120 seconds. "Strikingly, the effect of the low-frequency stimulus on the cochlea persists for longer than the duration of the stimulus itself," Drexl points out. Further experiments will probe the possibility that this phenomenon may be linked to noise-induced auditory damage, one of the most common causes of hearing impairment in industrialized countries."

So apart from the implications for industrial workers & those living near to a wind turbine, I believe it has implications for audio reproduction. I know from my own & others experience that reducing noise leads to a significant perceptual improvement & I'm not talking about noise that can be heard. The best example I can give is reducing or eliminating common mode noise (usually from a ground loop) results in a benefits across the sound stage - better separation of instruments, better placement, a more 3D soundstage, more dynamics - a generally more realistic illusion (if that's not an oxymoron )

The information about noise that I have considered significant & held onto:
- Jocko (an RF & audio clock expert) has consistently claimed that close-in phase noise was the most significant aspect in clocks for auditory perception
- a number of people, including Opus111 (who used to post here & goes under Abraxalito on DIYAudio) have reported the perceptual improvements in using a huge over-capacitance at the supply pins of audio chips (both digital & analogue). We are talking about 10,000 more than the usual 0.1uF bypass cap specified in datasheets for this pins.
- He also premised that noise modulation was the big issue with modern D-S DAC chips
- In my own experiments, although not targeting LF noise directly, I have found that reduction in noise sources have greatly enhanced auditory perception.

I have agreed with Opus's opinion that some re-evaluation of our perception influence of noise floors was needed & overdue.

This new research helps to bring together these strands & focus attention on what may be a crucial factor in our auditory perception. BTW, I saw this recently which gives an insight into the one role of noise in our auditory perception - Comodulation Masking release

Any thoughts?

LL21 said:

I am looking forward to reading your article and to the responses here...thanks for posting this. Interesting.

Click to expand...

Thanks!
Here's some points to ponder which might be relevant:
- this might begin to explain the "night & day" differences reported by audiophiles & yet the absence of measurements is cited as one element in the argument for why audiophiles are "fooling" themselves. I've never seen measurements of VLF presented for audio devices as it was considered below perception & because it's also a difficult measurement to take, requiring a long running FFT & long averaging.
- this might begin to explain why "night & day" differences reported by audiophiles & yet the absence of blind test positive results. No properly run blind tests that I know of has tested VLF & it's affect on auditory perception & I discount all blind tests run under the pseudo-science belief system.

Here's a thought that could turn out to be ironic - what if the hullabaloo about high-res sounding better than RB & yet all the measureists stating that we can't hear > 20KHz is actually to do with a difference in very low frequency (VLF) noise? In other words the relaxed digital reconstruction filters used in high-res produced less VLF noise - the opposite end of the spectrum to where they are focussed

One incongruous element that keeps niggling at me, however, is the apparent high level of VLF noise in vinyl playback, yet good vinyl playback is hard to beat & mostly sounds perceptibly more lifelike than digital. Maybe other factors of the noise are important - it's quality, modulation, spectrum?

LL21 said:

Layman's thoughts/experiences:

1. My Velodyne is shut off (40db rolloff) above 40hz. Switch it on or off, and the sound goes from 'everywhere' to just the space between the 2 speakers. And suddenly, you are very much more aware that you are in your room. The sub seems to overlay some kind of information that overrides the room's noise or something...hence the often-used description of the 'walls melting away'. As soon as its switched off, you get this 'let down' feeling where you are back in your own room...the travel to another venue is gone. That's just the best way i can describe it.

2. As for measurements of VLF, the more i read, the more i suspect we have not truly married all the measurement data and how the human ear functions.

We know some stuff...like 3db increase might double acoustic power sound...6db for voltage sound pressure...but due to human hearing which seems to compress parts of its mechanics in listening at louder levels (fletcher-munsen curves or something)...it takes 10db for a human to actually hear a doubling of volume.

Click to expand...

Interesting observations, LL.
Others might offer more insights than I but I know that perceptually low freqs give body & weight to what we hear - it's partially at least because of better localisation signals at low freqs giving more 3D impression of the sound stage & partially because the low freqs are where all the ambience information is stored so we get the impression of the room size of the recording venue - just as you reported - the soundstage outside of your speakers ?

DonH50 said:

Of course close-in phase noise is the greatest contributor to audio noise as that is what translates to LF energy.

Click to expand...

Hi Don,
I presume yo mean "measured" LF energy rather than "perceived"?

Adding vast amounts of capacitance needs some exploration IMO. Op-amps and most balanced audio circuits (the majority of low-level analog stages and ADC/DAC analog interfaces even if the output is not balanced) have very high PSRR and CMRR is excellent at very low frequencies. However, coupling through biasing and feedback networks could easily swamp the intrinsic PSRR/CMRR of such circuits.

Click to expand...

Usually I see significant differences in datasheets between the PSRR of the + & - rails - would this not imply they weren't balanced circuits? There are usually other pins than the supply pins that are also PS sensitive & datasheets don't show any PSRR values for these. Recently replaced a 10K resistor (acting as a CCS) on the IREF pin of a DAC chip with a CCS & the difference was "night & day".

I wonder how VLF noise perception ties in with our LF hearing roll-off (e.g. Fletcher-Munson loudness curves)?

Click to expand...

In what way are you thinking of this tie in?

One thing that can be extrapolated from the comodulation masking release I gave a link to above is that when the timing of two elements in the sound suggests to our auditory processing they should be grouped together as one auditory object, it strengthens (more than doubles) our perception of that object. In the link above we can't hear a tone frequency above a certain noise level but if we then introduce a modulation of that noise that coincides with the tone frequency, the tone now becomes audible in the presence of the same level of noise - it emerges from the noise. What this suggests to me is that when our auditory processes group elements of the sound together into auditory objects, it enhances our perception of that object - the object becomes more starkly etched sonically (which could translate into hearing a more dynamic, realistic version of the auditory object.
So a simple change in noise can snap a sound stage into place

Here's a quote from DIYA 2013 that I'm trying to follow up
"Hmmm, I have found an interesting way to change the perceived noise, though I am having troubles measuring it.
ARTA running averaging of 2000 or so trials is showing up differences in VLF noise, interestingly.
Subjectively the difference is sort of like reduced jitter/imd, with deeper solider bass and clearer mid/highs.
L/R and depth imaging markedly improved, overall more realistic sound but no FR changes measured."

This again reinforces what I was saying on my other thread about objectivists & measurements - they assume that perceived differences are either frequency, amplitude or timing & therefore if the measurements show no differences in these there is nothing to investigate except the "self-delusion" of the listener.

How many people would run 2000 trials of ARTA?

DonH50 said:

I am not sure why the distinction between measured and perceived in this case? Close-in phase noise on a clock leds to LF modulation so the correlation is clear. That is the noise most likely to be audible. It is also comprised of high-order (f^N) terms and so is the largest noise component as well. PN at a 100 kHz I would expect to be inaudible, at least for me since I can't hear that high... It could be translated to the audio band through any nonlinear (mixing, sampling) operation, natch.

Click to expand...

Yes, I understand any broadening of the tails either side of the central spike in a graph of jitter is a function of the close-in phase noise. What I was trying to ascertain was this - any noise below 20Hz is considered imperceptible through our auditory senses yet I believe that it affects our overall perception of the rest of the audible range in significant ways. I mentioned Jocko who claims that close-in phase noise as low as 0.1Hz is significant (this might be hubri though?)

A balanced circuit is with respect to the signal, not the power rails, at least any definition I have seen. I was thinking you had an engineering background? My apologies if I assumed too much (I usually get blasted whether I talk over or under folk on the 'net, a no-win scenario). In any event your observation is absolutely correct since vew few circuits present the same isolation from both rails. Fundamentally, and hand-wavingly, one rail is usually tied to the load (output) side and the other rail to the bias side of the circuit, so the effective noise coupling is completely different. But, usually very high in either case.

Click to expand...

Sorry, brain fart there

A 10k-ohm bias resistor is not "high-impedance" in most circuits so I am not surprised a CCS made a large difference. It will also change the noise profile since you go from resistor noise to that produced by an active device. I would expect to add shot and flicker noise in addition to thermal noise. The output noise depends upon how the reference resistor is used in the circuit., as does the PSRR. Unless you have a very clean supply I would not tie the resistor that establishes a DAC's reference current to the supply, though some designs use that approach to track out power supply noise so there is no hard rule that works everywhere.

Click to expand...

It was a R to ground so the ground of a single supply DAC chip

I have not read your links yet. If we are more perceptive to VLF noise than previously thought, something I do not know but would not really surprise me based on my own experience, then would that modify the equal loudness curves? That was what I had in mind, purely curiosity on my part.

Click to expand...

Yes, it might be interesting. I'm certainly of the opinion that JNDs should be established for more complex signals than simple pure tones as I think this can be misleading

Aside: Many years ago studies (mine and others) showed that carbon film resistors actually had lower noise than metal film. Metal film processing has improved quite a bit since that time, so I do not know if that is still true, and of course metal films can also be made more stable over PVT and time than most carbon films. But, I still recall working on a high-end preamp back in the 70's or 80's that had a bank of metal-film R's clearly visible looking in from the top. They weren't connected anywhere, the designer said they were too noisy in the circuit, but at the time metal film was a must-have by marketing so they threw in a few on a dummy board that was really there for shielding (just a ground plane shield).

Click to expand...

Tut, tut - audio chicanery

Orb said:

Slightly different context but Hifi News did a very interesting investigation quite a few years ago (initially started by Jim Lesurf) showing massive very low frequency jitter affecting some USB DACs, I would need to dig out the two articles and under what situation/products this was identified but it did correlate with subjective performance.
I appreciate this is not same as OP but still interesting in how very low frequency behaviour can still be of influence in products not expected to be.
Cheers
Orb

Click to expand...

Hi Orb
It's called the IQtest & I posted about it here last year

Yes, Jim showed measurements for a DACMAgic DAC compared to a Halide Bridge (asynch USB) - I believe the low freqs timing blips were mostly because the DACMagic was using adaptive USB & not for the reasons Jim gave at the time - that it was the loading on the PC

andromedaaudio said:

hearing 0,1 hz ???
Which loudspeaker or sub is gonna reproduce that , the best units have a resonanz freq of around 20 hz , if you look at the slopes of them hypotheticly 0,1 hz would be for example 50 db under the listening level or something like that plus they would distort/break up like hell

Click to expand...

That's the point - we're not talking about "hearing" 0.1Hz noise but rather hearing the effects of such noise. I suspect it is so close to DC that it has a modulating affect on the PS & this results in ground currents shifting reference points, as Don says.