An interesting argument against sampling frequencies above 96kHz

The white paper is over 5 years old. The blurb you quoted from Benchmark is over 7 year old. Benchmark now uses 192K. Things change.

The white paper is still on the Lavry website and so are the Benchmark comments. The burden of proof is on you to show they recanted their earlier positions. Show me where Lavry has changed his mind. His stance on this topic is widely known in the industry so if he changed his mind I would like some proof. The reason they offer them is because mastering studios need them when mixes get sent to them with 4x sample rates or engineers are asked to record at those sample rates by clients. Also, some engineers just like the sound - some might like the sound the distortion adds, which is totally valid.

You seem to have a religious belief regarding high sample rates. I can tell you, as a professional classical musician who also records, that there are far more critical issues than sampling above 96k. When I down sample properly and dither from 96/24 or 88.2/24 to 44.1/16, it is extremely difficult to hear the difference, if at all, let alone 96 vs 192. I've done blind listening tests with my professional musician friends and the results have been right around 50/50, the same as guessing. I think I can hear the difference when I know which one I am listening too, but when it is blind, suddenly I cannot tell any more. I prefer to focus on better mic placement, better mics, better converters (not necessarily meaning higher sampling rates), and more importantly - better acoustics.
 
You can only do so much by ear. You need hard measurements.
You can do everything by ear. The ear is tje only arbiter. PS Audio has bomb proof measurements and Audio Note's measurements look like a bomb hit them :)

The latter company can easily design bomb proof measurements but when they experiment ans listen they found that a different course sounded better even though the measurement didn't look as good.

Now they could design like Benchmark and just lool at the computer screen. Or they could listen and say man that sounds dreadful we better do something.

Of course it's always nice when you have somerhing that sounds nice AND measures great because you can usually win the forum debates becauase you have some graphs to put up that look good. When you try and defend SET or NOS CD players --- well it becomes like defending a religion. Jusr listen or just have faith. For this Atheist I get to see what that's like I suppose.
 
...
Having just covered the California Audio show in July i found it interesting that the three best sounding rooms all used non oversampling DA converters with tubes and no digital filters. 2 use no analog or digital filters and all are relative stone age in the numbers game. Even on Hi Res recordings NOT played natively at their 24/192 res, they STILL sounded better. I suspect due to much better analog output stages.
...

While related I don't think this is a NOS vs. OS discussion but rather how far do you need to go in increasing sampling frequency during recording and playback using actual (not interpolated) samples. Let's get real, while redbook can sound nice on a NOS DAC, playing true hi-res on the same NOS DAC is a completely different experience.
 
And iy is puzzling playing a true hi res recording on a hi res DAC and playing same file on NOS dac the fact that the latter STILL sounds better is rather interesting. Which is why I figure it comes down to non filters and the quality of the analog output stage.

I do like PS Audio though..but my money goes for NOS until I can be convinced otherwise - I have a drive filled with 24/192 music - that sounds better on my NOS DAC than on my DAC with 32/192 ESS Sabre. It kind of goes back to the OP wondering whether to play at a lower resolution - which is why I say go with what you hear.

I continue to look at high resolution designed for computer audio playback - and will continue to do so - but so far I have not been overly impressed by the sound - perhaps - again - it is their analog output stage that is the culprit and not the technology. I have certainly not heard all the Computer Audio DACs out there. It only takes one.
 
Last edited:
There are really few compromises when listening to 24/192 on a quality NOS DAC that can handle it natively as this is pretty much as close to the real thing as it gets, no wonder it sounds better than ESS with its number crunching and DS modulators. I don't think NOS vs. OS argument exist at these sampling frequencies.
 
The distortion is measurable (see benchmark quote above) so why use sample rates above 96k?
Similarly, virtually every amplifier produces lower distortion at 16 ohms than at either 8 ohms or 4 ohms. Why use speakers with impedances lower than 16 ohms? How about yours?

Everything in audio reflects a balance. The "measurable" difference with the PCM1794As used in my DAC is 0.0007%. Yes, that's seven ten thousandths of one percent. Do you believe you can detect that difference?

Also, your quote must reflect a different DAC than mine given this comment about clock speed: "The 4x (176.4 kHz and 192 kHz) mode..." Let's contrast that with what TI says in the datasheet about the PCM1794A:

"PCM1794A 24-Bit, 192-kHz Sampling, Advanced Segment, Audio Stereo Digital-to-Analog Converter...System Clock: 128, 192, 256, 384, 512, or 768 fS"

Obviously,
TI designed it for use at 192 kHz where 192 is only 1.5x clock speed.

Higher sample rates allow the designer to use even more gradual and less damaging filters in the playback process. Having said that, I am definitely one who believes in diminishing returns and when given the option between 96 and 192, I usually choose the lower cost and less space taken with 96. But most certainly NOT because I believe the 192 version would sound *worse*.
 
Similarly, virtually every amplifier produces lower distortion at 16 ohms than at either 8 ohms or 4 ohms. Why use speakers with impedances lower than 16 ohms? How about yours?

Everything in audio reflects a balance. The "measurable" difference with the PCM1794As used in my DAC is 0.0007%. Yes, that's seven ten thousandths of one percent. Do you believe you can detect that difference?

Also, your quote must reflect a different DAC than mine given this comment about clock speed: "The 4x (176.4 kHz and 192 kHz) mode..." Let's contrast that with what TI says in the datasheet about the PCM1794A:

"PCM1794A 24-Bit, 192-kHz Sampling, Advanced Segment, Audio Stereo Digital-to-Analog Converter...System Clock: 128, 192, 256, 384, 512, or 768 fS"

Obviously,
TI designed it for use at 192 kHz where 192 is only 1.5x clock speed.

Higher sample rates allow the designer to use even more gradual and less damaging filters in the playback process. Having said that, I am definitely one who believes in diminishing returns and when given the option between 96 and 192, I usually choose the lower cost and less space taken with 96. But most certainly NOT because I believe the 192 version would sound *worse*.

According to Dan Lavery, a sample rate of 60k is plenty high enough to deign the filter. He should know.
 
According to Dan Lavery, a sample rate of 60k is plenty high enough to deign the filter. He should know.
He is welcome to his opinion - as different as it is from many DAC designers. And I would most certainly not choose using OPA134 op amps in the analog stage as he does with the DA2002. It's fine for the $500 Music Hall unit in my garage system, but not for an $11k "high end" unit. He should know better.

My listening experience is different.
 
Last edited:
You can do everything by ear. The ear is tje only arbiter. PS Audio has bomb proof measurements and Audio Note's measurements look like a bomb hit them :)

The latter company can easily design bomb proof measurements but when they experiment ans listen they found that a different course sounded better even though the measurement didn't look as good.

Now they could design like Benchmark and just lool at the computer screen. Or they could listen and say man that sounds dreadful we better do something.

Of course it's always nice when you have somerhing that sounds nice AND measures great because you can usually win the forum debates becauase you have some graphs to put up that look good. When you try and defend SET or NOS CD players --- well it becomes like defending a religion. Jusr listen or just have faith. For this Atheist I get to see what that's like I suppose.

What authority do you have to say that designing high end converters can be done without any measurements? Do you know the ins and outs of designing an ADC and DAC?
 
He is welcome to his opinion - as different as it is from many DAC designers.

My listening experience is different.

Have you done a double blind listening test? If you expect something to sound better, your brain will often confirm its own bias.
 
Have you done a double blind listening test? If you expect something to sound better, your brain will often confirm its own bias.
I have compared a wide range of recordings and arrive at the same conclusion as other trusted ears.
 
I have compared a wide range of recordings and arrive at the same conclusion as other trusted ears.

So the answer is no then.

You have no way to know if you can hear it or not with enough double blind tests.

I’ve been doing informal tests with my professional musician friends for years and the results are 50/50, same as guessing. And I’d say they have pretty good ears.
 
Sure, but coming at it from my perspective, as a professional classical musician who also does recording, I want to see the specs and I want to listen.

And coming from my perspective a trained listening professional:beatnik:, if that is even a thing:banana:, 23yrs as a Passive Sonar Tech, I was teaching baby sailors how to listen critically to ocean back ground noise and find the man made sounds when I retired in 2007.

I guess my “Why?” should have been why get caught up in specs and measurements. And trust me I am a spec whore just like most.

It’s been my personal experience with 10-20 people every week in a class for 3 years that specs only point you in the right direction, at the end of the day they don’t mean diddly squat (ha, first time I have got to use that). It’s what each person hears. I can measure all kinds of pieces parts from a submarine on a time/frequency display but you’ll never ever hear it.

Of course like most things audio it’s just MHO.

Edit; and as a Sonar Tech I maintained my own gear (we had some sweet 1” tape machines) and have been working on Marine Gas Turbine and Power System controls for the last 10yrs so I understand all the measureing and tech stuff also

:beerchug:
 
Last edited:
The white paper is over 5 years old. The blurb you quoted from Benchmark is over 7 year old. Benchmark now uses 192K. Things change.

What’s your point, technology hasn’t changed much over the last 5 yrs... oh wait ummm
 
And coming from my perspective a trained listening professional:beatnik:, if that is even a thing:banana:, 23yrs as a Passive Sonar Tech, I was teaching baby sailors how to listen critically to ocean back ground noise and find the man made sounds when I retired in 2007.

I guess my “Why?” should have been why get caught up in specs and measurements. And trust me I am a spec whore just like most.

It’s been my personal experience with 10-20 people every week in a class for 3 years that specs only point you in the right direction, at the end of the day they don’t mean diddly squat (ha, first time I have got to use that). It’s what each person hears. I can measure all kinds of pieces parts from a submarine on a time/frequency display but you’ll never ever hear it.

Of course like most things audio it’s just MHO.

:beerchug:

All good points, but let’s keep in mind the point here; I’m not interested as intersected in what people think they can hear but what happens when that claim is subject to double blind tests, as the Boston Audio Society did in an extensive test, which found that professional recording engineers on numerous very high end systems could not hear the difference between 44.1 and 96.

Also, the data is there that converters perform worse at ultra high sample rates than 96.
 
Back
Top Bottom