What does this actually do?

[bennetng]

So are we saying that with a correctly band limited signal we will recreate the original signal correctly?

If "correctly" means "transparent" then my criteria is pretty loose and a Realtek will pass, but if people do measurements and zoom in everything as close as possible then ultimately differences will show up.
https://hydrogenaud.io/index.php/topic,114871.0.html

Some DACs have excessive high frequency roll off and a upsampler can solve the issue.
http://archimago.blogspot.com/2015/08/measurements-ponoplayer-another-mans.html

Also this:
https://hydrogenaud.io/index.php/topic,110058.0.html

I have a feeling that some DAC designers deliberately use some potentially nontransparent filters in order to convince their customers that their products are good enough to demonstrate the benefits of hi-rez.

 

[March Audio]

By correctly I mean in accordance with sampling theorem.

 

[bennetng]

Then you need to have a target. For example, how many dB do you need to attenuate at 22.05kHz to justify the correctness? Would you consider 1dB drop at 21.5kHz and 60dB drop at 22.05kHz as "fail"? Some people like Rob Watts prefer very close to ideal reconstruction, some people like Bob Stuart prefer the opposite but both parties don't offer reliable explanations to their preferences.

The remaining designs for example may be able to achieve 85-95% of flat passband response and some degrees of aliasing/imaging suppression. When talking about music production, apart from ADC people may often put their recorded signal through samplers, resamplers, effect plugins and so on and all of them have their own filters with different performance.

If an ADC drops something like 120dB at 20kHz with 44.1kHz sample rate then a lot of resamplers or DACs can reproduce the digitized signal without problems, but what if an ADC or resampler is flat up to 20kHz and drops 30dB at 22.05kHz? The Sampling Theorem doesn't address how real world DAC/ADC or any hardware/software based processing are performed.

 

[Blumlein 88]

Then you need to have a target. For example, how many dB do you need to attenuate at 22.05kHz to justify the correctness? Would you consider 1dB drop at 21.5kHz and 60dB drop at 22.05kHz as "fail"? Some people like Rob Watts prefer very close to ideal reconstruction, some people like Bob Stuart prefer the opposite but both parties don't offer reliable explanations to their preferences.

The remaining designs for example may be able to achieve 85-95% of flat passband response and some degrees of aliasing/imaging suppression. When talking about music production, apart from ADC people may often put their recorded signal through samplers, resamplers, effect plugins and so on and all of them have their own filters with different performance.

If an ADC drops something like 120dB at 20kHz with 44.1kHz sample rate then a lot of resamplers or DACs can reproduce the digitized signal without problems, but what if an ADC or resampler is flat up to 20kHz and drops 30dB at 22.05kHz? The Sampling Theorem doesn't address how real world DAC/ADC or any hardware/software based processing are performed.

You get aliasing and imaging. How much is too much? Not that hard to define.

 

[March Audio]

Then you need to have a target. For example, how many dB do you need to attenuate at 22.05kHz to justify the correctness? Would you consider 1dB drop at 21.5kHz and 60dB drop at 22.05kHz as "fail"? Some people like Rob Watts prefer very close to ideal reconstruction, some people like Bob Stuart prefer the opposite but both parties don't offer reliable explanations to their preferences.

The remaining designs for example may be able to achieve 85-95% of flat passband response and some degrees of aliasing/imaging suppression. When talking about music production, apart from ADC people may often put their recorded signal through samplers, resamplers, effect plugins and so on and all of them have their own filters with different performance.

If an ADC drops something like 120dB at 20kHz with 44.1kHz sample rate then a lot of resamplers or DACs can reproduce the digitized signal without problems, but what if an ADC or resampler is flat up to 20kHz and drops 30dB at 22.05kHz? The Sampling Theorem doesn't address how real world DAC/ADC or any hardware/software based processing are performed.

I think you may be missing the point I am alluding to. The crux of the issue in this context is not the replay image filter but the anti alias adc filter.

 

[bennetng]

I think you may be missing the point I am alluding to. The crux of the issue in this context is not the replay image filter but the anti alias adc filter.

Okay. Let's say an ADC is flat up to 21kHz and over 100dB of attenuation at 22.05kHz, if it is played by a DAC which is flat up to 20kHz and drops 6dB at 21kHz then would you consider it a perfect reproduction? This question doesn't involve transparency.

 

[andreasmaaan]

Okay. Let's say an ADC is flat up to 21kHz and over 100dB of attenuation at 22.05kHz, if it is played by a DAC which is flat up to 20kHz and drops 6dB at 21kHz then would you consider it a perfect reproduction? This question doesn't involve transparency.

My question is, what's gained by having a perfect (as opposed to merely transparent) reproduction?

 

[bennetng]

My question is, what's gained by having a perfect (as opposed to merely transparent) reproduction?

Read reply #41 and #42
I only need transparency so I will let someone else to answer this question.

 

[sergeauckland]

My question is, what's gained by having a perfect (as opposed to merely transparent) reproduction?

Brownie points in the magazines and on Forums, so one sells more.

It may or may not make one iota of difference in a blind test, but commercial sales aren't done on blind tests.

S.

 

[March Audio]

Okay. Let's say an ADC is flat up to 21kHz and over 100dB of attenuation at 22.05kHz, if it is played by a DAC which is flat up to 20kHz and drops 6dB at 21kHz then would you consider it a perfect reproduction? This question doesn't involve transparency.

we are talking about the transient/timing response of the DAC in this context. Is the anti image replay filter irrelevant if the ADC is being done correctly.

 

[bennetng]

we are talking about the transient/timing response of the DAC in this context. Is the anti image replay filter irrelevant if the ADC is being done correctly.

Then the "we" doesn't include me.

 

[Purité Audio]

Rob Watts’ PowerPoint presentation from London’d ‘Canjam’.
https://www.head-fi.org/threads/hug...e-official-thread.885042/page-2#post-14383243
Keith

 

[bennetng]

Rob Watts’ PowerPoint presentation from London’d ‘Canjam’.
https://www.head-fi.org/threads/hug...e-official-thread.885042/page-2#post-14383243
Keith

 

[Dialectic]

Not to get off topic, but here is how brilliant technical mind and writer Rafe Arnott attempted to explain different ways of doing the processing to accomplish upsampling in a comment to his ad copy--er, report--regarding the Hugo M Scaler:

AFAIK dCS use a bespoke PLL (phase-loop lock) with a variable frequency oscillator to accomplish their upsampling... but don't quote me. I've put an email in to dCS this morning on this, so as soon as I get confirmation I'll post it.

Meridian is a different story, I don't have a direct contact there yet, but their Ultra DAC "... is also the first product in the world to include MQA Limited’s Hierarchical Converter Technology, which employs multiple converters to increase temporal resolution while reducing noise and quantization errors."

I don't know if that means they're using an FPGA or a PLL, but I'll see if I can get someone from there to get back to me.

I can see how this guy replaced Lavorgna, but how does he publish for the same media entity that publishes, among other enlightened folks, Professor Rubinson?

Crazy.

 

[derp1n]

Some discussion of this device and an upsampling listening test going on over at Computer Audiophile.

 

[bennetng]

Some discussion of this device and an upsampling listening test going on over at Computer Audiophile.

Niccceeee!
IIRC mansr is SoX's maintainer, so it is a war between freeware and the WTF filter.
In fact I am highly impressed with SoX, it can even resample to and from prime numbers with great speed and accuracy.

 

[sarscott]

Hard to judge Rob Watts if you have not had extended experience with his products. The Chord M Scaler is not a scaler you can casually hear for a few seconds and come to a conclusion about the sound quality. This is exactly why Rob Watts doesn't believe in double blind testing as the sample sizes are too small/too short and people's memories are fallible. Audio products can sound great for short listening periods where the brain compensates for the errors in reproduction but then quickly become irritating and/or grating as the brain becomes fatigued. Listening to music this way is work and not pleasurable. This is exactly why the CD format has failed as the reproduction was mostly prone to a lot of errors that the brain can easily detect as the digital sound is unnatural.

Comparing the work that Rob Watts is doing to washing a dish for the 101st time is not accurate at all as the brain is very difficult to deceive when the sound is full of non-random/artificial artifacts and/or noise.

Overall audio reproduction has improved greatly over the last 10 years in general but still much improvement is needed.

If you have the chance, go to a dealership that has the Dave DAC and a direct connect to full range crossover-less speakers and listen for an extended period over a week at different times with your favorite music and I guarantee you will be sold on Rob Watts work. The proof is in the experience.

 

[SIY]

Hard to judge Rob Watts if you have not had extended experience with his products. The Chord M Scaler is not a scaler you can casually hear for a few seconds and come to a conclusion about the sound quality. This is exactly why Rob Watts doesn't believe in double blind testing as the sample sizes are too small/too short and people's memories are fallible.

IOW, the same excuses other parity product manufacturers use. And inaccurate excuses at that.

This is exactly why the CD format has failed

Yes, 30 years of market dominance is a failure. Right. And of course, the format has been supplanted by downloads because of the obsolescence of physical media in general rather than by any general concerns about quality of the medium.

On the surface, this looks like the usual marketing blather. Do you have any actual data to support your assertions?

 

[sonci]

View attachment 14274

This Guy is a genius! While other remove or add features from the same product justifing price, this one increase Taps.

 

[JJB70]

The sad thing is that I do believe that Chord are a very capable company and can design excellent products when they want to. Unfortunately they are also happy to jump on the bandwagon of selling overpriced audio jewellery. They can also sell stuff with sub-standard batteries then shoot the customer if they complain, but that's another story.