• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

High Resolution Audio: Does It Matter?

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Here’s Stagetec’s explanation:

https://www.gearslutz.com/board/att...ital-other-gs-faux-pas-truematch-function.pdf

I guess measurements would decide if it’s BS or real.
Yes, looking at some patents. Exactly as I expected. Gain ranging preamplifier prior to conversion with multiple converters. I would look into the difference between dynamic range, and SNR.

Our ears have maybe 60 db SNR, and the muscle that adjusts sensitivity allow a 120 db dynamic range. So they are doing something similar. They likely have 110 or 120 SNR, and with multiple ADC's and preamp gain riding maybe they manage 150 db dynamic range.

So misdirection. Maybe in a sense 150 db dynamic range, but not the full resolution of 150 db they are implying. And that's if they really achieve what they claim.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
Gain ranging is done most commonly on measuring instruments, as can be seen in the 'staircase'noise plots that result. Thats' perfectly OK, but it's 'misdirection' at best and dishonest at worse in an audio product where gain ranging can't be used in a live recording situation.

S
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
I also have some microphones with self noise of 3 and 4 dbSPL equivalent, but I've never gotten close to that in any live recording situation.

The ones I've tested accomplish this through slack diaphragms which means limitations at high SPL. You can also do it with large diaphragms, but now we're talking significant coloration. TANSTAAFL

Reality is, 15-20dB SPL is good enough for music recording. Like you say, that noise is swamped by the room and the people in it.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,385
Location
Seattle Area
Here’s Stagetec’s explanation:

https://www.gearslutz.com/board/att...ital-other-gs-faux-pas-truematch-function.pdf

I guess measurements would decide if it’s BS or real.
Very interesting. Thanks for posting that. Here is the key figure and text there:

1532280317895.png


This is indeed what my Audio Precision analyzer does.

In the case of my analyzer though, switching gain stages creates a glitch. This shows up as a sudden variation in the measurements which is not consequential. In an ADC however for music these glitches must not exist. They say they have logic to avoid this but would be fascinating to measure a signal that goes just above and below the threshold for two ADCs and see the effect.

I would also worry about noise modulation/stepping due to different gain stages getting turned on and off with different noise levels.

Still, the technical in theory is sound.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
The problem with most such schemes is that signals don't usually appear in the real world as single tones. Thus they cannot accurately resolve a signal at full-scale in the presence of another at or near the noise floor. And if you are trying to gain-range the converter on the fly, dynamically following the signal amplitude in real time (sample-by-sample), designing and aligning switching gain stages and aligning signals is challenging (/understatement -- you need to align in amplitude and time, without adding switching glitches or anything else, pretty tough!). Plus you have to deal with saturation of the lower-signal (higher-gain) paths. In my past this been exactly the problem faced: full-scale, or nearly full-scale, signals in the presence of neighboring signal at the limit of the system's dynamic range. Having a number of signals at the ADC's input ranging from say 0 dBm to -150 dBm can make life interesting.

That said I have seen some schemes that do a good job, though none at say the -100 dBFS level. But I have not been involved with those sorts of things for a few years now.
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
The ones I've tested accomplish this through slack diaphragms which means limitations at high SPL. You can also do it with large diaphragms, but now we're talking significant coloration. TANSTAAFL

Reality is, 15-20dB SPL is good enough for music recording. Like you say, that noise is swamped by the room and the people in it.

Those I have are Lewitts in 1 inch diaphragm. They do have somewhat reduced max loudness. 138 without padding as I recall. You can pad them 20 db without hurting the mic or incurring distortion so they aren't too bad on that front. I also have a couple Shure KSM44a which have a max spl of 152 db at .5% distortion. Also a 1 inch diaphragm. Both of these are multi-pattern dual diaphragm microphones. The SDCs on hand do have self noise in the teens. I've never found that noise to be a problem.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
That is how I understood Truematch to work. And is the difference in SNR and DR. The SNR would likely be 110 db or so if good. Yes between quiet signals and loud signals the DR of such a system might be 28 bit, but that doesn't mean it will resolve to 28 bits in the presence of a loud signal. If the varying loudness were a big limiting factor that might help, and in a mixing I can imagine places it would have benefit. As a straight recording you aren't running out of range as it is. I usually set the average level at about -18 db with a few peaks above that and basic noise isn't an issue already.

It also occurs to me if you do your mixing in a DAW instead of a console, most everything is done at 64 bit, so moving levels and such between tracks is not anything to be bothered about.
 
Last edited:

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,068
Likes
16,598
Location
Central Fl

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,758
Location
My kitchen or my listening room.
Lots to parse there, and frankly until proven otherwise I don't believe them.

I have a recording interface which claims 129 db dynamic range. Now it is very quiet and very wide range. Yet to get to 129 db you are basically allowing a quiet gain circuit to offer the ability to record low level signals with little noise added from the gain circuit, and comparing that to higher level signals without the gain. Not exactly something you can actually do for recording. It does have an excellent 118 db or so of dynamic range at any given gain setting for the ADC. I believe the Truematch system is doing something similar. So some misdirection involved there I am pretty sure.

I also have some microphones with self noise of 3 and 4 dbSPL equivalent, but I've never gotten close to that in any live recording situation.

Interesting. Air noise at the ear is about 6dB SPL (white noise) 20 to 20k.
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,758
Location
My kitchen or my listening room.
Yes, looking at some patents. Exactly as I expected. Gain ranging preamplifier prior to conversion with multiple converters. I would look into the difference between dynamic range, and SNR.

Our ears have maybe 60 db SNR, and the muscle that adjusts sensitivity allow a 120 db dynamic range. So they are doing something similar. They likely have 110 or 120 SNR, and with multiple ADC's and preamp gain riding maybe they manage 150 db dynamic range.

So misdirection. Maybe in a sense 150 db dynamic range, but not the full resolution of 150 db they are implying. And that's if they really achieve what they claim.

Actually, you can argue for 80dB of adjumstment in the cochlea, although after 60dB the mechanisms start to widen the filter bandwidths substantially. The stapes reflex kicks in above that, but really doesn't get you that much more. So I'd still go for total of 90dB in anything approximating a clean 'reception'.

Some things to remember.

1) at 120dB there is measurable nonlinearity in air transmission
2) at 140dB there is mostly nonlinearity
3) 194dB SPL is 1 atmosphere peak sounds, and doesn't exist because the whole linear transmission approximation fails utterly at such levels.
4) The air noise at the eardrum is between 6dB SPL and 8.5dB SPL. This is due to the brownian motion of the air molecules hitting the eardrum. Going below that over 20-20k is an "interesting" thing to accomplish.
5) In order to get a an SNR of some particular level, you need to convert the SNR back to an absolute ratio, then square it, to find out how many electrons/second need to be moving in that circuit to accomplish that kind of SNR. The charge on the electron is not as small as imagined.

Some points I make now and then:

1) a 32 bit fixed-point convertor, set to range from atmospheric self-noise to peak, provides a peak level of about 198dB SPL. The only possible term for that is "military" and the only "transducers" tend to consist of unstable chemicals that only work once per device.

2) For 144dB (24 bits) you need 2^48th electrons per second flowing at peak level from your microphone capsule/whatever into the preamp. Do that math, ok? At 600 ohms, thats'a bout 25 milivolts out of the mike.

3) for 32 bits, you need 2^64th electrons per second. Now that is NOT happening in any safe environment. Nope.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Interesting. Air noise at the ear is about 6dB SPL (white noise) 20 to 20k.

Those mikes are 3-4 db rated with an A-weighting. So they aren't that without the weighting. I would expect at least 3 db difference in the unweighted result if not close to 6 db.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
Also, I should point out that the ear has an SNR of about 30dB, but the loudness compensation processes in the cochlea map that over about 80 dB or so, so that at any level, you can resolve about 30dB below that level, give or take.
That 30dB SNR is very much in line with my experience. Under programme conditions, especially with low dynamic range music such as most pop, a 40dB S/N ratio is perfectly adequate. FM radio rarely provides more than a 60dB S/N ratio except under ideal conditions, and with wide dynamic range music such as BBC Radio 3, no noise is audible in the background. The 80+ dB S/N ratio of CD is entirely adequate under pretty much any domestic condition, even one as quiet a rural location as I'm in, so struggle to understand why 24 bit is bothered with, except as a recording standard where editing and other manipulations will be done. Internally, production mixers work to 32 or 48 bit (possibly more now) but that's so the maths can be done in DSP. Any other reason is just marketing.

S.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Ok, nice read. But what is the conclusion here, does HiRes make a difference in our ears or not?
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,440
Likes
632
Ok, nice read. But what is the conclusion here, does HiRes make a difference in our ears or not?
Again, the best summary of the meaningful science on that question is the Joshua Reiss meta analysis here:

http://www.aes.org/e-lib/browse.cfm?elib=18296

There is no simple answer or any indication that it makes an obvious, slam dunk difference. Different test studies with different methodologies have reached different conclusions or levels of significance. Some listeners in some studies heard a difference fairly consistently, some did not. Some test studies got overall higher levels of discrimination among all trials, some did not. Was it the differing methodology of the test studies, perhaps?

I think perhaps the most interesting aspect of the paper is the seeming dependence on prior training in how to listen and what to listen for in the various tests. Harman also considers this very important in their speaker testing. Many test studies for hirez discrimination did not do that. Those that did had generally higher average levels of discrimination. But, the fact that hirez seems to require this for better discrimination indicates the subtlety of any differences in listening. Even so, no test study achieved above 70-odd percent discrimination.

So, read Reiss’ paper as you will. At least it moves well beyond the widely ballyhooed Meyer-Moran study, which was hopelessly flawed and useless, as Amir agrees.

I am in general agreement with Reiss. The evidence suggests the difference is at best slight and not always detectable. I may be deluding myself, but I like to believe that with experience, I have learned to be able to hear the small difference it makes at least a fair bit of the time. It happens that my listening is primarily in Mch, which is most always recorded and delivered in hirez. Personally, I like it and prefer it. But, I make no claims about huge differences from hirez.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,168
Likes
3,714
and, from another reputable corner, Dan Lavry...back in *2004*

Nyquist pointed out that the sampling rate needs only to exceed twice the signal bandwidth.
What is the audio bandwidth? Research shows that musical instruments may produce energy
above 20 KHz, but there is little sound energy at above 40KHz. Most microphones do not pick
up sound at much over 20KHz. Human hearing rarely exceeds 20KHz, and certainly does not
reach 40KHz. The above suggests that 88.2 or 96KHz would be overkill. In fact all the
objections regarding audio sampling at 44.1KHz, (including the arguments relating to pre
ringing of an FIR filter) are long gone by increasing sampling to about 60KHz.



I lost count long ago how many times I've linked hi rez cheerleaders to this paper (and others of his where he fleshes his argument out). Lavry has a stake in this -- he makes highly respected, certainly 'audiophile' grade, ADCs. And yet the dominant engineer-side voice re: hi rez has long been the increasingly quackery-friendly Bob Stuart of Meridian.
 
Last edited:

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,168
Likes
3,714
So, read Reiss’ paper as you will. At least it moves well beyond the widely ballyhooed Meyer-Moran study, which was hopelessly flawed and useless, as Amir agrees.

Reiss's choice of data to crucially include or exclude in his meta-analysis is hardly flawless either. And one of the major so-called flaws of M-M is merely a laughably transparent case of goalpost-shifting on the part of hi-rez advocates.

As for the central question 'does it sound different': if it takes a meta-analysis to suggest a difference at all, *you* , the random listener, probably can't hear it.
 
Top Bottom