Upsampling 16/44.1 collection a good idea?

[MaxwellsEq]

My summary:

1. The designers and manufacturers of DACs know how they perform each processing step, but it's unlikely that they will let us access that information without NDAs. Based on data sheets, subject matter experts can make excellent informed hypotheses about what each part does.

Because of this, only the designers/manufacturers know for certain whether their method of sample-clocking has fewer or lesser artefacts than doing it in software externally ahead of feeding the DAC. But they are unlikely to reveal this information (except under NDA).

For the rest of us, we have to depend on measurements of the black-box DAC as implemented by manufacturers of commercial products. From these tests (such as done by Amir), it's hard to see how software-first conversions can create benefits which are audible.

 

[Brian Hall]

My summary:

1. The designers and manufacturers of DACs know how they perform each processing step, but it's unlikely that they will let us access that information without NDAs. Based on data sheets, subject matter experts can make excellent informed hypotheses about what each part does.

Because of this, only the designers/manufacturers know for certain whether their method of sample-clocking has fewer or lesser artefacts than doing it in software externally ahead of feeding the DAC. But they are unlikely to reveal this information (except under NDA).

For the rest of us, we have to depend on measurements of the black-box DAC as implemented by manufacturers of commercial products. From these tests (such as done by Amir), it's hard to see how software-first conversions can create benefits which are audible.

Based on tests it is obvious that extra upsampling is useless and probably even counterproductive.

 

[sunjam]

My summary:

1. The designers and manufacturers of DACs know how they perform each processing step, but it's unlikely that they will let us access that information without NDAs. Based on data sheets, subject matter experts can make excellent informed hypotheses about what each part does.

Because of this, only the designers/manufacturers know for certain whether their method of sample-clocking has fewer or lesser artefacts than doing it in software externally ahead of feeding the DAC. But they are unlikely to reveal this information (except under NDA).

For the rest of us, we have to depend on measurements of the black-box DAC as implemented by manufacturers of commercial products. From these tests (such as done by Amir), it's hard to see how software-first conversions can create benefits which are audible.

100% agree with you.

The key point here is "hard to see how software-first conversions can create benefits which are audible" (statement 1, factual) does not automatically imply "there is no audible difference from the sofware-first conversions created" (statement 2).

Statement 1 and statement 2 are not equivalent. Anyone who claim statement 1 implies statement 2 would base on wrong logical reasoning. It is a mistake we can make easily.

All I can see is that "there is no evidence to prove there is audible difference from the sofware-first conversions created"

You can also view it this way: "there is no evidence to prove there is no audible difference from the sofware-first conversions created"

It means there is no conclusive result yet. We need to do more testing, measurements, etc... if we want to find a definite answer for the question like "is there any evidence to prove there is audible different from the audio signal reconstructed by using a modern DAC with and without external upsampling".

i.e.:

There is no evidence to prove there is audible difference when using an external up-sampling s/w. (I think most of us agreed here)
Meanwhile, there is also no evidence to prove there is no auidble difference when using an external up-sampling s/w.

Food for thought,

1. do we believe that there exists better advanced upsampling algo than the one built-in a DAC chip?
2. do we believe that a PC (e.g. modern i7, 32G RAM + Nvidia GPU) would have more processing power than a DAC chip (e.g. ESS9039Pro <== one of the most advanced chip availalbe in the market)?
3. if we have better algo and better computing power, does it mean the end result would be better?

I don't any objective answers (as I have no proof) but my own personal view are Yes, Yes, and Yes.

 

[danadam]

It means there is no conclusive result yet. We need to do more testing, measurements, etc... if we want to find a definite answer for the question like "is there any evidence to prove there is audible different from the audio signal reconstructed by using a modern DAC with and without external upsampling".

And how will you decide that the definite answer was found?

Consider two scenarios:

Scenario 1: the definite answer is "yes".
In that scenario you keep looking and eventually, after a year or 100 years or 1000 years, you find the proof that there's is an audible difference. Cool, you've found the answer, case closed (possibly by your great-great-grandchildren).

Scenario 2: the definite answer is "no".
In that scenario you keep looking but you don't find any proof. When will you stop and decide that you've found the answer and it is a "no"?

 

[BDWoody]

You can also view it this way: "there is no evidence to prove there is no audible difference from the sofware-first conversions created"

Ok, we've had enough of a Tempest in Russell's Teapot for a while. I think a little break to do more of your own research might be great.

 

[pkane]

Looks like, these days. ASR has developed a lot of group think over time which hardened up the red lines and made the group decision what's beyond one of those red lines and what is not very binary and non-debatable. Crustification at work. <-- strictly IHMO, of course.

ASR is still full of valuable information and of course Amir's review work is outstanding as it creates a data pool for comparisons like we never had before.

I agree, there's a certain part of ASR membership that just jumps into action to defend the perceived red line whenever some newcomer comes and tries to challenge it. I see it more as an ASR self-defense mechanism than anything more sinister. There's too much groupthink on the outside of ASR, where nothing is ever challenged, and any challenge is automatically dismissed "because your system isn't resolving enough or you are deaf".

The opposite of what happens on ASR happens daily on many other fora. The direct outcome of this inability to think critically are products like PGGB, for example (directly applicable to this thread). PGGB is the direct result of lack of critical thinking. The author claims the error of -600dB or better for his reconstruction algorithm. Tell me, Klaus, would you spend $1500 on a software package that off-line converts your entire library to DSD1024 over a period of weeks, requiring a PC that needs hundreds of Gigabytes of memory, 20+ cores CPU, GPU, and tons of storage? Well, there are many that are buying it. In fact, they hear major audible improvements, even compared to the already major audible improvement that were claimed previously by HQPlayer users.

Heck, some actually started running their library through BOTH, PGGB and then through HQPlayer for an even "better" result In my view, it's the direct outcome of audiophiles not being challenged to think for themselves, believing marketing or the larger groupthink that permeates that world of "everything matters," and that "science can't know everything," and that only their flawed perception can be the judge.

Critical thinking is a skill, and must be learned. I see less and less of it being taught in schools. Unfortunately, an internet forum is not a place to undo the damage caused by the primary education system -- most of us here are not teachers, and it's not our job to re-educate others. I'd much rather see newcomers be challenged to use critical thinking, to spend time investigating and learning, rather than automatically assuming that just because "I heard it" it must be so. But I understand why this is not so after many dozens of new users each week start posting the same exact boring, unsubstantiated ideas and thoughts that have not been subjected to even an ounce of critical thinking.

 

[Purité Audio]

Couple of ‘Hitchhikers guide to the Galaxy’ quotes on the front page surely it can’t be all that bad!
I use to occasionally see Douglas Adams in ‘Subjective Audio’ in Mornington Crescent way back when.
Keith

 

[pkane]

Couple of ‘Hitchhikers guide to the Galaxy’ quotes on the front page surely it can’t be all that bad!
I use to occasionally see Douglas Adams in ‘Subjective Audio’ in Mornington Crescent way back when.
Keith

It's actually a technical marvel of a product. Completely useless for the main claimed benefit of audible improvements, but then, internally it uses 256 bits precision for calculations, runs slow, costs a lot, and requires a major investment in time and equipment to operate. High cost, extreme complexity and inconvenience, all backed up by extreme claims. This is what drives a large part of the audiophile consumer market.

 

[Purité Audio]

Sounds ideal as a vinyl replacement activity.
I am reminded of the time that a friend visited with his ‘XXX Hi-End ‘ software, don’t get excited it was just playback software.
We spent nearly forty minutes fiddling with it but couldn’t get it to emit one note of music.
Keith

 

[pkane]

Sounds ideal as a vinyl replacement activity.
I am reminded of the time that a friend visited with his ‘XXX Hi-End ‘ software, don’t get excited it was just playback software.
We spent nearly forty minutes fiddling with it but couldn’t get it to emit one note of music.
Keith

I've had many debates with Peter, the author of XXX Hi-End in the past. He even sent me one of his (complex!) USB cables to test at one point. He wasn't too happy when I found that it produced no measurable effect, except for some additional noise in some configurations. I did try his player, too. It was a bit clunky to configure, but it did work for me. Can't say that it was doing something special, but it did run and play music, though the interface was particularly ugly, in my opinion

 

[Purité Audio]

Did you ever try his ‘H’ shaped dac which always reminded me of. ’Rimmer’ from Red Dwarf .
Keith

 

[pkane]

Did you ever try his ‘H’ shaped dac which always reminded me of. ’Rimmer’ from Red Dwarf .
Keith

I've seen it and heard from others (and Peter) about it, but have never tested one. The claim was that it was more immune to noise generated by the PC by virtue of better isolation and low CPU usage when used with the XXX Hi-end player. Peter was trying to prove that even the act of moving a mouse on the PC would cause noise to be added for more mainstream DACs by the extra CPU utilization, while his DAC was immune.

 

[KSTR]

@pkane, you certainly have good points in your large post above and I do agree pretty much on all of it.

 

[KSTR]

PGGB is the direct result of lack of critical thinking. The author claims the error of -600dB or better for his reconstruction algorithm.

Just had a closer look at PGGB and yes, this is completely insane, throwing quad-double (256bit) math at simple audio problems. A solution in search of a problem.

For the fun of it and as I already have my own time-domain convolver in double precision, I've used a brickwall filter kernel made with rePhase for a test run of a 2x upsampler. With that filter kernel (Albrecht-11 term, -6dB @ 21500Hz for 88.2kHz target sample rate, 2048 taps, -300dB down at fs/2 and above) -- and not using any dithering for that test -- I could achieve sample identity for a full-scale signal in the 24bit integer domain for any frequency up to 21100Hz, that is, the upsampling contains all the original sample values except for an ultra narrow range at the top end. For music tracks, with much lower HF content, the match would be even better. As such, it can't get any better than this, me thinks.... and it's already overkill ;-)