[B] NEED HELP - Diganosing audio differences between source components [/B] - Page 10

It was in my posts.


I assumed nothing. I showed you AES paper from Fielder which had actual measurements of live performances and analysis of the audio chain to preserve the same. You substituted a gut feeling for that kind of research. Sadly, you have to read what is written to have realized that and you don't see to want to do that.


They will have more to chew on soon .

You mean this:


I'll be interested to see if you can wring another 50db of dynamic range out of it. According to Fielder's paper, the appropriate adjustment from peak to rms appears to be 3db:

"The peak acoustic level was 129db SPL, or its 126db SPL rms equivalent..."

Accordingly, it looks like the 70db rms value should be increased to 73db peak. You've still got 47db to go.

And frankly, if Keith Johnson can't achieve 120db dynamic range in a 176.4/24 file, I don't know who can.


Didn't see any AES papers or other research showing that the "creative types" in general have the necessary skill sets to achieve a recording with 120db dynamic range. That's the part you just assumed, which is surprising given what you think of their ability to remember not to truncate when converting 24 bit to 16 bit.


Okay. I'll be enjoying some CDs in the interim, even though I won't know what their dynamic range is or whether the mastering engineer applied dither when he bounced the recordings down to 16 bits. I might even put on an LP and try not to have a nervous breakdown over its lack of 120db dynamic range.

I finally got around to reading the Fielder paper on dynamic range. I'll quote my own previous post below to give a point of reference.


It turns out that Fielder does do this. The number he comes up with for threshold of audibility for white noise is 3.8 dB SPL. His studies of peak SPL of live performances gives a maximum value of 129 dB SPL. The sine wave having this peak SPL has an average SPL 3 dB less than this, or 126 dB SPL. So the required dynamic range he comes up with for this case is 126-4 = 122 dB. In Fielder's words:


So, what assumptions does he make about the reproduction system? Here is the block diagram he uses.



One might wonder where the analog volume control after the DAC is, and why he omits it. Here is the reason, in Fielder's own words.


This doesn't correspond with the reality of current HT pre/pros and receivers though, nor with audiophile-style systems having a separate DAC and preamp either. Virtually all HT pre/pros and receivers have digitally-controlled analog volume control chips after the DAC, such as the Cirrus CS3318 found in the higher-end Onkyo receivers and pre/pros.

So if his 90-100 dB dynamic range claim for analog preamplification is correct, this eliminates any possible noise improvement from having more than 16 bits in the digital system.

The Cirrus 3318 is about the best of this category of gain-control chips. Cirrus claims a minimum 121 dB dynamic range for the CS3318, but they do not specify any relevant test conditions. This is surely a best-case number, with the attenuation ahead of the internal op-amp set to 0 dB and the gain of the internal op-amp also at 0 dB, with output voltage at maximum, where it will overdrive any normal power amplifier (about 7.5 Volts peak). Adjusting this for a more normal 2VRMS output degrades that number by about 8.5 dB to 112.5 dB, but that still assumes no input attenuation and no op-amp gain. With any real-world attenuation, you're back to 16-bit noise performance.

And of course, these numbers assume the only noise contribution is the analog volume control, when in fact there are noise contributions from microphones, mic preamps, mixers, A/D and D/A converters, and power amps as well. Fielder looks at these contributions individually and does not address what occurs when they are all combined. This gives an overly optimistic outlook.

So it appears that the practical dynamic range is going to be limited by the analog circuits in the chain, making recordings having greater than 16 bits unnecessary for all but the most extremely optimized playback systems and recordings.

You are correct in general. Audition has a standard feature for doing this conversion via a check box. It adjusts the appropriate numbers by 3 dB.


I used the peak reference when I did my analysis, so this adjustment (as meaningless as it is given the size of the numbers involved) is not necessary because it was built into the results I presented.


Exactly. I've done a few test recordings in a larger, modern purpose-designed venue (about 800 seats, adjustable acoustical treatments for the type of presentation whether choir, band, or drama) with no audience or musicians present, HVAC off and actually obtained electrical (including the microphone internal electronics) performance on the order of 100 dB. Of course, when I enabled the mics acoustically, the available dynamic range went back down to less than 70 dB.

BTW, the "Samson and Delilah" HDtracks recording that Amir brought to the discussion does have an approximate 3 second lead out (right before the digitial zeroes that Amir tried to include in his analysis) where it appears that just the residual sound of the hall has been recorded. Its average level is about 45 dB below the peak of the loudest sound on the recording. Spectral analysis shows its largest peak at about 15 Hz (environmental), followed by one at 50 Hz (power line related - it is an European venue) and it does show evidence of noise shaping.

In addition to eliminating from consideration the analog preamplifier at the playback stage, while he spends a lot of time discussing the dynamic range of microphones, and touches on the analog-to-digital converter, his analysis is completely silent on the dynamic range of the analog microphone preamplifiers as well.

So he did all of his measurements without a mic pre-amp? Maybe it is best to have him say and show it:

AES Paper:
Human Auditory Capabilities and Their
Consequences an Digital-Audio Converter Design
LOUIS D. FIELDER
Dolby Laboratories, Inc., San Francisco, California 94103

"Figure 11 is the spectral comparison of hearing acuity with examples of microphone, home listening room, recording studio, and symphony hall noises. All but the recording studio spectra came from measurements by the author, while that spectrum was presented in a study of broadcasting studio noise by Meares and Lansdowne. Although these microphones, home listening room, and recording studio noises represent quiet examples of noise environments of their type, they are no means exceptional. Comparison of the four specra to the hearing acuity shows that the microphone and recording studio noise has little effect on the perception of distortion and noise since they are comparable in level to hearing acuity."

He goes to show how industry measurements of mic noise/dynamic range (even when it includes a-weighting) is completely flawed since it ignores psychoacoustics. Here is Figure 11:



The graph shows once again the fallacy of using single dB numbers to describe the noise floor of a system/room/equipment/music. Such a number does not represent how we actually hear sounds. You have to make narrowband measurements, apply psychoacoustics to them and then compare them to the threshold of hearing one segment at a time. Our hearing has a very non-linear response with most sensitivity in the mid-band. It is noise that exists in that narrow region that is of most importance. You can have a ton of noise in lower and higher regions and it may not be audible.

So believe in what you want to believe. But please don't put words in man's mouth, saying he is not competent enough to use and analyze how a microphone works relative to its noise performance. The man specializes in this field and knows way more than all of us .

Even though you and I have had this discussion over two threads now, lasting many days, it is disappointing that you are still ignoring psychoacoustics on audibility of noise by constantly trusting a number shown by a meter, devoid of what what two ears+brain hear. More below.


Let's look at the threshold of audibility:



We see that at 20 Hz (second bar to the left) our hearing sensitivity is 80 dB lower than at 3 Khz. 15 Hz will likely be even lower. At 50 Hz, it is 50 db lower. Therefore these frequencies are much less interesting to look at after we compensate for how inaudible they are. I doubt that if you were in that haul you would be hearing 15 Hz rumbling away.

Let's investigate this further using your method. Assuming you are right that the beginning of the track represents the actual noise of the venue, I went ahead and ran a spectrum analysis for that region alone:



We see that you are indeed right that a lot of the noise is in low frequencies which thankfully we don't hear that well per above. The noise at the critical, most audible range of 3 Khz is in -108 dB or 54 dB lower than the peak in low frequencies! While one would need to perform the analysis of Fielder/Stuart do to arrive at audibility of noise relative to tones in the threshold graphs, we can safely conclude that there is no way this room only has 45 dB or whatever you are using for this recording.

So as you see, we completely invalidate the notion of using a meter to determine the noise floor of the system. You cannot be spectrum blind when analyzing noise. This is on top of the fact that we hear through the noise so even that wouldn't set the limit anyway.

And no, I don't agree that is noise shaping. This is a 24 bit track. Why would it have noise shaping? And why would they stuff quantization noise in the low frequencies alone when they have plenty of room in the ultrasonics? As Fielder measurements show, and one knows from science of noise reduction, low frequency noise is not attenuated easily and gets through from many sources. Fortunately if it is not audible, it is not important.

Bottom line, we can second guess this work all we want. But if we start with throwing out how we hear, it is not a proper defense. This analysis is properly done by Fielder/Stuart across many research papers and if it were bogus, it would not get that far. The work paints a credible picture of best recordings having 120 dB of range that should be preserved.

Did you know that there is a difference between a microphone and a microphone preamplifier? It's like the difference between a phono cartridge and a phono preamplifer. The microphone preamplifier provides gain, generally 60-80db. It does not do so without adding noise. Talking about the dynamic range of a microphone or a phono cartridge tells you nothing about the dynamic range of a microphone preamplifier or a phono preamplifier.

I've read the paper you referred to. It contains passing references to microphones but no discussion of microphone preamplifiers, which would fall under the term "recording equipment" he uses in his paper, but does not discuss.

In the dynamic range paper we are discussing in this thread, he talks about the dynamic range of several excellent brands and models of microphones - Schoeps, Neumann, Sennheiser, Bruel & Kjer. His concern with the dynamic range of microphones is whether they will overload at high SPLs. The ones he discusses do not, provided they are not placed too close to percussion instruments in multi-mike setups. So for the purposes of the paper, they are not a limiting factor in recording 120dbs of dynamic range.

Curiously, there is no discussion of microphone preamplifiers and their noise contribution.

Don't worry about me. It seems you think Fielder tried to pull a fast one with those articles/research, performing *measurements* with microphones without said pre-amps. How exactly do you pull that off? How about this text in the paper you say you have read?

"The next section consists of the microphones, the microphone amplifiers, and any mixing console before the recorder...."
Doesn't it look like is has heard of microphone amps? He goes to say:

"Several other microphones were also measured to ensure
that the performance represented by the microphone in Figure 3 was comparable to
other existing microphones. The results are shown in Figure 4. Four different microphones were measured which had overload levels between 120 to 140 decibels. They
were all condenser microphones and as the graph shows the noise levels in the
3 - 7 kHz region were witin 5 dB of each other. In summary, it is shown that close
talking techniques and the proper selection of a microphone produces no limitation
or reduction on the dynamic range requirement as determined by the playback experimerits."

This all reads like someone not using a microphone plus recording chain to analyze the problem? Of course not. All of his data is driven from measurement with microphones. How else would he accomplish this task?

As I said, the man's work is thorough and proper. Layman semantic "gotcha" is not a proper counter and just adds noise to the conversation, pun intended . You have some data and research that says mics can't record more than 60 to 80 dB with their pre-amp included with perceptual model applied to them? That is what is needed. Not the continued reliance on single figures of dB pulled out of thin air no matter how many times I explain that is a meaningless value.

The noise referred to in the text, and reflected in the graph, is not the electrical noise of the microphone preamplifiers but the acoustic self-noise of the microphones themselves. Fielder is very clear about this:

"Dynamic-range reduction due to microphone imperfections is determined by an examination of the best examples of microphone design. To this end, four excellent recording microphones and one ultralow-noise measurement microphone, all of the condenser type, were examined for maximum undistorted sound level reproduction and equivalent acoustic self-noise...

A further examination of microphone technology is warranted. This begins by comparing the noise of the Bruel and Kjaer 4179 to the recording microphones and observing that its noise level is substantially lower than the others. This is true because its design lowered the primary source of self-noise, the diaphragm damping element, as shown by Tarnow [34]. The design of the 4179 microphone, as discussed by Frederiksen [35], reduces the value of the noise that induces diaphragm damping and equalizes the resulting resonant rise in the frequency response. This method could also be applied to the design of a recording microphone, allowing lower noise levels. If the noise levels of the 4179 are scaled to the diaphragm sizes of 12—18 mm for recording microphones, the resultant microphone noise would still be at least 5 dB below the hearing threshold."

There is no discussion of the electrical noise of the microphone preamplifiers.

Amir, Amir, Amir.

Here's what I wrote:


The operative words being "Lead out", which means in digital audio recording terminology means the end of the track.

http://en.wikipedia.org/wiki/Optical_disc_authoring

"The lead-out area is the ending part of the CD session."

When someone shows you measurements of a hall and the text clearly says the chain includes that of recording, it is assumed. But since you like to see it in black and white from him, here you go:

Determining NoiseCriteria for Recording Environments*
ELIZABETH A. COHEN**, AESMember
Charles M. Salter Associate, San Francisco, CA 94104, USA
AND
LOUIS D. FIELDER, AESFellow
Dolby Laboratories, Inc., San Francisco, CA 94103, USA

"… subsequent measurements were made with the Bruel and Kjaer ultra low-noise system consisting of a 1-in 4179 microphone and 2660 pre-amplifier. Its extremely low self-noise levels are showin in Fig. 5 "



Here is the noise measurements of consumer listening rooms:



If a microphone+its pre-amp in a real consumer room had lower combined electrical and acoustic noise than the threshold of hearing, there is nothing else to be debated. We can achieve such a noise floor. Here is the conclusion of the paper you quoted:

"When a representative sampling of classical music recording situations was taken, a dynamic-range requirement between 90 and 118 dB was obtained for listeners with good hearing. Most situations had a dynamic-range requirements exceeding the 98-dB capacity of the 16-bit linear PCM recorder without preemphasis."

It doesn't get more black and white than this. To say that he managed to get this paper published in the Journal of AES with its "peer review" and such yet he had forgotten to include the noise from the mic pre-amp is just silly. But if you want to believe it, that's cool.

Arny, Arny, Arny, the noise is the noise. It makes no difference if it is the silence at the end or beginning. Here is the "lead out" to make you happy



This time I circled the more critical region around 3 KHz. We see all the same points I made in my previous post. We are still at -108 dB in the 3K area. And far higher in low frequencies which would fool a meter showing a "dumb" SPL value devoid of psychoacoustics.

By the way, since this is showing true hall noise, it was not fade out region. Therefore, your method of cutting this out of analysis by Audition to determine the dynamic range was incorrect! That analysis showed 111 db of RMS differential which is pretty close to this rough number.

This is all fine and good, but Fielder's cherry-picked collection of microphones and prognostications about what could be done to create low noise recording microphones, is all moot. Nobody cares. There never was a new generation of mainstream recording microphones with at or below hearing threshold noise levels.

Nobody cares. Nobody cares because of the 500 pound gorilla in the room - room noise.

Amir, don't confuse the fact that I now know that I have to take you though audio 101 a step at a time with the individual steps.

So the first step was to get you looking at the same part of the same audio track that I was (ironically, based on your choice of tracks which is poor when it comes to proving your point).

Now, I have to teach you how to analyze noise levels. The rookie mistake we are concentrating on today is the same one I just corrected JA for when we were talking about amplifier SNR. Note that he had the good sense to stop arguing that point with me. You didn't. ;-)

The mistake I corrected JA for is using a FFT to establish what a noise floor is. This fallacy is based on the fact that when it comes to random signals, a FFT will give a lower noise floor the more points you put into the analysis.

The correct way to measure noise in a frequency range is to filter the noise so that you are measuring the noise in just the desired frequency range, and then measure the output of the filter. This is for example how we use things like "A weighting": First we filter the noise with a pre-defined filter and then we measure the output of the filter.

When working with Audition we can easily measure noise correctly because Audition has a incredible set of filters that we can use to synthesize just about any filter we want.

Looking at the Fletcher Munson curve:



We see that human hearing reaches its peak sensitivity in the range from 2.5-4.5 KHz. If I apply what Audition calls a "scientific filter" with Butterworth characteristic and 4th order roll-offs with these points, and then measure the results I get the following numbers:

Left Right
Min Sample Value: -21.24 -14.81
Max Sample Value: 22.78 17.95
Peak Amplitude: -63.16 dB -65.23 dB
Possibly Clipped: 0 0
DC Offset: 0 0
Minimum RMS Power: -94 dB -90.38 dB
Maximum RMS Power: -81.81 dB -81.72 dB
Average RMS Power: -90.15 dB -87.82 dB
Total RMS Power: -89.92 dB -87.73 dB
Actual Bit Depth: 32 Bits 32 Bits

Using RMS Window of 20 ms

Bottom line - Still nothing that can't be handled with properly noise shaped 16 bit quantization.

The Bruel and Kjaer 2660 microphone preamplifier is the dedicated 20db gain preamplifier specifically designed for the 4179 sound measurement microphone. The 4179 will not work without the 2660. They even use proprietary connectors. That is why he refers to them as a "system". The 2660 is not a suitable microphone preamplifier for recording microphones like the Neumann, Schoeps, and Sennheiser he discusses.

You are in denial. He did not discuss the electrical noise of the recording microphone preamplifiers just like he eliminated from consideration the analog preamplifier in the playback system. Continued hand waving won't change that.

Again, check with Bruce Brown at Puget Sound Studios. I believe he, like many recording engineers, is a fan of the Manley microphone preampiliers. They have an SNR of 80db. Why would anyone use a microphone preamplifier with an SNR of 80db you may ask. The answer is simple: they sound great, which is a more important criteria than absolute dynamic range.

AFAIK, the Manly mic preamps, being tubed, aren't likely to have SOTA dynamic range. If Manly makes it, and its SS, they brand it a Langevin...

I'm not going to do a study of their complete line, but the one I'm looking at called a "MICMaid" says 118 dB, when SOTA for SS is about 10 dB better (ca. 128 dB).

However, you have to take the dynamic range specs for mic preamps with a grain of salt, as they are usually based on the largest signal they can handle without clipping with the gain turned all the way down, as compared to the equivalent input noise with the gain turned all the way up. People don't use them that way - they adjust the gain to suit the session and pretty much leave it alone for the session. Real world is closer to 100 dB which is just fine.

The problem with distant miking of classical music is that you are at the mercy of environmental noise. The graph at http://www.stereophile.com/content/low-noise and included in this posting shows the noisefloor of a recent piano project where I had the luxury of both a very quiet hall and a staff who let me take an afternoon analyzing and eliminating noise sources. The bulk of the noise energy is at low frequencies where the ear is less sensitive.

For reference, the green trace shows the noisefloor of a perfect 16-bit system analyzed with the same FFT settings for the recording's noisefloor. You are welcome to draw your own conclusion.

John Atkinson
Editor, Stereophile

That is correct, Mr. Krueger. I meant to write -140dB. The Levinson amplifier’s noisefloor is lower than that level despite the high signal level. The noise has not risen, thus preserving this amplifier’s very high dynamic range capability.


Not without limit, otherwise you would get the ridiculous situation that increasing the FFT size to infinity would give you an infinitely low noisefloor


Thank you for the link. I am familiar with the subject of FFT processing gain. The graph I referred you to was taken with a 16,384-point FFT with AP's Equiripple window and 32 averages to reduce the granularity of the noise floor. Repeating the test with a 32,768-point FFT would reduce the apparent noisefloor by 3dB, which I believe indicates that the Levinson’s level of noise is not the graph’s limiting factor.

Incidentally, if you notch out the fundamental and harmonics and calculate the RMS sum of the remaining bins, the indicated S/N ratio would be independent of FFT size, the 3dB drop of each bin when you double the FFT size being compensated for by the fact that there are now twice as many bins to be summed.

John Atkinson
Editor, Stereophile

Should the listeners pay extra for such capability? The capability that's reserved for an apocalyptic audio event that may happen once in every 1000 years or so?

Seeing Halley's comet in 1986 was an unforgettable experience. Not having been around in 1910 and unlikely to be around in 2061, it was a once in a lifetime event. Not sure I would have paid $8500 (the cost of the Mark Levinson No. 532H) to see it though, but some would. Depends on your interests, priorities and financial means. Maybe someone wants to pay $8500, plus the order of magnitude extra cost of everything else that is required, to hear a recording with 120db dynamic range, on the off chance that such a recording is actually ever made.

Whether they want it is a matter of personal choice, but the fact is that if they want that dynamic range capability, it will, of necessity, be expensive. All the extremely quiet, extremely powerful amplifiers I have measured cost a lot of money.

John Atkinson
Editor, Stereophile

You shouldn't have bothered. You said you performed a certain analysis and I did the same thing which actually confirmed what you said, except that your conclusions from it were incorrect. You said and I quote, " the "Samson and Delilah" HDtracks recording that Amir brought to the discussion does have an approximate 3 second lead out (right before the digitial zeroes that Amir tried to include in his analysis) where it appears that just the residual sound of the hall has been recorded. Its average level is about 45 dB below the peak of the loudest sound on the recording. Spectral analysis shows its largest peak at about 15 Hz (environmental), followed by one at 50 Hz (power line related - it is an European venue) and it does show evidence of noise shaping."

You said that portion has the residual hall noise and certain frequency peaks. I did the same analysis using the same software you had. As I said, i showed the same things you said. Except that your read and interpretation was wrong:

1. There was no noise shaping as the signal was 24 bits. Wasn't that a digital audio 101 mistake to assume such?

2. You ignored psychoacoustics that tells us low frequency noise is much less audible. You proceeded to tell us what the peak at 15 Hz was in amplitude and left it at that.

3. I went on to say that your analysis is not correct anyway with the comment, "While one would need to perform the analysis of Fielder/Stuart do to arrive at audibility of noise relative to tones in the threshold graphs..." But since you ran with it, I showed that analyzing the signal at high level points to you drawing incorrect conclusions from it.


Why don't you provide the spectrum so that we can finish this cat and mouse game of "you don't use the same segment?"


Whatever is wrong here per above, was your idea. You went to do a spectrum analysis and I did the same thing. That aside, you are mistaken to assume that if you keep increasing the number of points the noise floor keeps going down as if to give you a lower value than the signal itself. FFT can add its own noise. It doesn't subtract from system noise.


That is only a subset of what you have to do to measure the *audibility* of the noise. You have to then subject it to psychoacoustics to convert the audibility of noise to tones used in the red Fletech-Munson curves. This is exactly what Fielder/Stuart had done and rock_bottom talked about and confirmed after reading the articles. This is why I noted your analysis was incomplete here but since you did it, I ran with it to show you that you were providing a misleading view of the noise spectrum regardless.


That analysis is useless. You need to use ERB analysis which takes into account the bandwidth of our hearing system which changes at different frequencies. All of this is properly done by Fielder and your back of the envelop crude approximations are not it. And the conclusions reached by him using proper analysis and real data arrives at the conclusions I have post from him.

Bottom line: you thought you could spin the data from this track to make a point and it backfired because using the same analysis as you did, it pointed out to completely different conclusions which actually proves the soundness of Fielder methodology. Only now you have gotten religion to filter the noise and make it narrowband. How come all along you have been giving us single number dB values if spectrum is important? I am glad we at least got you to understand you can't look at composite numbers a meter shows. Maybe in the future you will appreciate the importance of psychoacoustics on top of that related to auditory bandwidth of the ear.

Seems when the shoe is on the other foot and you have to defend a spec, psychoacoustics and how we hear is the first thing that is thrown out. Not once but repeatedly. You say you have read the fielder paper where he goes through why mic specs provided by manufacturers is completely wrong because it doesn't take into account how we hear. The analysis he does does that. Here it is again:


You see how he plots the effective audibility of the mic noise relative our hearing threshold and how its "dB" number varies based on frequency? In your next post, can you explain why you are refusing to accept this major point?

And that "500 poud gorilla" was characterized using proper science of how we hear and this was the results:



And here are performance venues:



In the last post, you said you believe (now) that noise analysis has to be filtered to the appropriate set of frequencies. That is what he has done and then some. He shows that we can at both end of this chain achieve inaudibility targets. Yes, it costs money and effort to get there. But such a performance can be achieved. Your statistics that it can't is due to faulty methodology and analysis of the same triggered by ignoring how we hear and how our our ears do not have flat sensitivity to noise.

Should we filter out all the music in the same manner as you propose with respect to noise. If, from a psychoacoustics viewpoint, we can't hear the low-level noise falling within certain frequency bands, then we can't hear the low-level music falling within those frequency bands either. Problem is that most of the fundamental tones of musical instruments fall below the frequency range where the ear is most sensitive. Might affect the dynamic range requirements.

JA,

Am I reading your graph correctly as showing that if we apply noise shaping when we dither down from 24 to 16 bits, we should be able to achieve all the dynamic range required to play back your wide dynamic range recording.

Also, would you care to address Fielder's comment that with an analog preamplifier in the playback chain, the best we can hope for is dynamic range of 90-100db.

Why? Someone made that music using their ear. The "filtering" was done for you .


Per above, a human made the levels as they saw fit for you to hear subjectively. The case of noise here is people using a meter and then trying to say that is what we hear. The music was not mixed and approved by a meter.

Both of the plotted microphones are sound measurement microphones and their required dedicated preamplifiers. And even one of the two shows audible noise in the range of 10db in the region where the ear is most sensitive.

Here is what he said with respect to recording microphones:

"Present recording microphones have audible dynamic ranges between 108 and 115 dB and are shown to be significant limiting factors in the creation of noise-free recordings at natural levels."

And that excludes consideration of their associated recording microphone preamplifiers.

I have yet to see anything which shows that dynamic range requirements cannot be handled by a 24 bit recording bounced down to 16 bits using appropriate noise shaped dither.

And then we still have the problem of the analog preamplifier in the playback chain which Fielder explicitly states limits dynamic range to 90-100db.

That's right: 108 to 115. Well in excess of flat dithered 16 bit.


That's you imagining that you can measure a microphone without a pre-amplifier.


I have yet to see you demonstrate a few tracks that have noise shaping. I asked this question before and you just ignored it. Nor have you shown any data that the entire or even most of the industry does that. I post how they are cagey and unclear in application of dither and noise shaping.

In absence of the industry performing noise shaping, the best path is for them to give us the 24-bit samples. There is no doubt then.


That is not what he says. He says, "the disadvantage with this situation is that dynamic range of most analog control preamplifiers is limited to 90-100 dB..." See the word you eliminated?

Today there are DACs that have high resolution digital volume controls and do not have this limitation anyway.

You continue to put spin on the data in this research, reading stuff into it that the conclusions of the articles clearly dispute. Here they are again as I post before:


And on how quiet the bottom of the dynamic range is:


Lots of defenses are put forward but at the end of the day, they are devoid of proper analysis performed in those papers.