Upsampling 16/44.1 collection a good idea?

[dlovesmusic]

Playing around with NOS DACs and DSD upsampling instead of speakers and room correction is like trying to move a mountain with a fork when you have an excavator right in front of you.

Don’t think playing around with external upsampling and speakers and room correction have to be mutually exclusive tho.

My listening room is pretty adequately treated acoustically and run audiolense convolution filter to incorporate subs with my main speakers and actually find the room treatment and room correction helps me easier to tell the difference between switching of digital reconstruction filters on the fly…

 

[pkane]

Don’t think playing around with external upsampling and speakers and room correction have to be mutually exclusive tho.

My listening room is pretty adequately treated acoustically and run audiolense convolution filter to incorporate subs with my main speakers and actually find the room treatment and room correction helps me easier to tell the difference between switching of digital reconstruction filters on the fly…

Sure, there are plenty poorly designed filters out there. Most are targeting audiophiles. A proper reconstruction filter is not a secret and is not difficult, and that’s all one needs in a DAC. In terms of impact on realism of sound reproduction, speakers and room correction are light years ahead of tweaking DAC filters.

 

[dlovesmusic]

Sure, there are plenty poorly designed filters out there. Most are targeting audiophiles. A proper reconstruction filter is not a secret and is not difficult, and that’s all one needs in a DAC. In terms of impact on realism of sound reproduction, speakers and room correction are light years ahead of tweaking DAC filters.

I am listening to the new Taylor Swift album as we speak, many of the tracks have quite a bit of strong transient and ringing (intentional or not), swapping to a shorter length minimum phase filter with steeper roll-off and attenuation, I thought the change is pretty obvious for the better…

 

[dlovesmusic]

I understand that here on asr ppl generally do not like to talk about things that are not proven objectively by measurements. And I personally don’t believe in things like cables, audiophile ethernet switch, fuses and other junks.

But this topic of external upsampling/ downsampling and application of digital reconstruction filter is something I do have doubts on as much as no solid evidence present re: it makes a difference or not

 

[Mnyb]

I am listening to the new Taylor Swift album as we speak, many of the tracks have quite a bit of strong transient and ringing (intentional or not), swapping to a shorter length minimum phase filter with steeper roll-off and attenuation, I thought the change is pretty obvious for the better…

Hmm but wich one is a proper reconstruction of the song if audible in a proper unsigthed test ?

With elaborate setups as yours there are probably some pitfalls ?

 

[Mnyb]

If you hear some audible difference but the measurements tell you that they should be the same, how would you interpret the result?

One way to interpret the result is that:

1. You brain just fool you, they are indeed the same in reality (if you consider the measurements taken are correct)
2. They are indeed not the same (if you consider the measurements taken are not correct, something affected the measurements)

What if... just what if:

both the measurements are accurate/precise and there are indeed audilbe difference?

Could they co-exist? If yes, what else we are missing here?

Human bias thats the factor in almost all these cases otherwise you can measure the difference. Filters actually do something .

 

[Mnyb]

Do we have something defined as a reference so that we can tell if a reconstruction is proper?


Food for thought: just wondering is it possible that the measurements that we are taken is not comprehensive enough to cover all the "audible difference”

That’s why amir measures DAC’s pkane can probably describe what a proper reconstruction filter is supposed to do.
Sadly for marketing reasons many DAC’s let’s you pick several different ones .

That’s a Russell’s teapot argument people have tried for decades. You can not prove negatives no one found Santa yet ( he simply hides somewhere else ).

Bertrand Russell’s once postulated that there is an teapot in orbit in space and challenged people to prove him wrong.

Elaborate software setups has some pitfalls like handling of intersample overs enough marginal to 0dB so that the EQ filters does not clip reconstruction filters with so wrong parameters that they do alter the fr response etc etc . That would be me reading the future in tealeafs to further try to get what’s up with Taylor swift I leave that discussion I don’t have a clue on all things involved.

 

[dlovesmusic]

Hmm but wich one is a proper reconstruction of the song if audible in a proper unsigthed test ?

With elaborate setups as yours there are probably some pitfalls ?

In this case with a modern day recording, the mixing and ringing are possibly intentional.

But there are cases for some of the older, worse recordings like some of the digitized analog recordings - rock stuff from the 70s/80s, this could actually be a solution to an issue

 

[Mnyb]

In this case with a modern day recording, the mixing and ringing are possibly intentional.

But there are cases for some of the older, worse recordings like some of the digitized analog recordings - rock stuff from the 70s/80s, this could actually be a solution to an issue

Suppose a Taylor Swift song can “break the rules” and contain samples that are “illegal” within the given sample rate of the recording ? But then all bets are off the song can not be properly reconstructed due to it being broken and you pick from an assortment of different bad behaviours and pick the least offensive

 

[dlovesmusic]

Suppose a Taylor Swift song can “break the rules” and contain samples that are “illegal” within the given sample rate of the recording ? But then all bets are off the song can not be properly reconstructed due to it being broken and you pick from an assortment of different bad behaviours and pick the least offensive

Sorry I am not sure if I understand your comment. I think there is not necessarily anything wrong with that album, it just so happen the music itself has some strong transient and ringing, some of these effects can be reduced with the application of certain minimum phase digital reconstruction filter

 

[antcollinet]

"Better Music" = "More Attractive Music"

But surely for that to be a definition of better music (Ie a characteristic of the music - rather than a characteristic of you), it would have to be universal.

Ie more attractive to everyone - or at least a vast majoity.

Otherwise you are just saying “I prefer it” which has 0 value as a statement to anyone else.

 

[pkane]

I am listening to the new Taylor Swift album as we speak, many of the tracks have quite a bit of strong transient and ringing (intentional or not), swapping to a shorter length minimum phase filter with steeper roll-off and attenuation, I thought the change is pretty obvious for the better…

Haha! Is there a really a filter that makes Taylor Swift sound “better”??? You should look into proper listening testing practices.

 

[antcollinet]

My intention is to exchange ideas and help me (and hopefully others) to learn

How does learning about your preference for something that is objectively inaudible help anyone? It can only mislead them.

 

[dlovesmusic]

Haha! Is there a really a filter that makes Taylor Swift sound “better”??? You should look into proper listening testing practices.

Haha, now we are into the realm of subjective impression and personal preferences when it comes to TS!!! lol

I previously had a brief conversation with archimago on this topic and I do believe a little steep high end roll-off could be good for harsh pop/rock and the minimum phase filter will remove some of the pre-ringing from poorly low-passed content (like loud or clipped modern pop/rock)...

 

[MaxwellsEq]

According to the scientific studies performed by Dr. Floyd E. Toole it is, at least to the majority of people who participated in the studies. If you have not read any of his studies, I recommend you do so. They were enlightening to me.

EDIT: Here is a passage from p. 11 of Floyd E. Toole's book Sound Reproduction (3rd ed., 2018):

Long ago, respected audio engineer John Eargle (1973) recognized the consequences of mismatches between monitoring and home systems in what is done during the recording process. Borja (1977) noted "dramatic and easily audible" differences in recorded sound quality caused by recording engineers compensating for spectral defects in control room monitor loudspeakers. He found that the inverse of the frequency response of the monitor loudspeakers could be seen in the spectra of the recordings. If the monitor loudspeaker had a response peak, the recording exhibited a spectrum dip - the engineer simply did what was necessary to make it sound right in the control room. This is not a new problem. Neither has the problem gone away. In an interview reported by Gardiner (2010) UK producer Alan Moulder, discussing the popular Yamaha NS-10M small monitor (see Section 12.5.1), said that "if you don't do anything they sound kinda boxy. They definitely make you work hard to make things sound right. You have to carve a lot out frequency-wise to make a track sound hi-fi." So, the process is to use a flawed loudspeaker and then equalize the mix to sound "hi-fi" to the mixer. Why? This mix can then only sound similar if reproduced through similarly flawed loudspeakers. This product and its distorted spectrum became so accepted by recording engineers that at at least one modern - fundamentally neutral - monitor loudspeaker has built in a switchable equalization to replicate it (http://barefootsound.com/technology) where they describe: "The 'Old School' setting captures the essence of the ubiquitous NS-10M, rolling off the sub-bass and top-end information, and bringing forward mid-range presence." So, the aberrant sound of the discontinued NS-10M may never go away.

I've read them, thank you. You are missing the point. 1) there's a reference; 2 there's a preference. They are not the same things.

 

[danadam]

swapping to a shorter length minimum phase filter with steeper roll-off and attenuation,

Maybe I have the nomenclature wrong, but doesn't "shorter-length" mean fewer coefficients, which implies shallower and earlier roll-off?

I thought the change is pretty obvious for the better…

Assuming that you actually can hear the difference, that very well may be, for you. However your preference can't really be a basis for claiming that this filter is proper or better (in general) than the standard one.

I do believe a little steep high end roll-off could be good for harsh pop/rock

Seems like a normal EQ would provide more control over this.

and the minimum phase filter will remove some of the pre-ringing from poorly low-passed content (like loud or clipped modern pop/rock)

Not sure what "poorly low-passed" means. And is the audibility of this pre-ringing just a belief or do you have something to substantiate it?

 

[MaxwellsEq]

They are perfect analogies because they work very similarly, read my explaination again.

They are poor analogies:
1. An amplifier and speakers can easily exceed the frequency of Red Book fs/2, but the best optical lenses are struggling to get close to or are worse (e.g. at the edge) than the pixel-per-mm/2 of a camera chip. This would be like CD comfortably getting to 20kHz but amplifiers and speakers all rolling off at 12kHz. This is the reverse model. You can NOT compare audio top end with fine edges in a picture!
2. Motion-picture technology limitations exist in the steps in vectors between separate samples. This has NO analogy in audio

 

[Tell]

Let me further explain the meaning of preference, IMO, with the help of the same picture again:

View attachment 364767

I made it smaller this time.

Let me state clearly first. All the following are my own personal opinion.

All the 3 pictures are from the same source (in my analogy, it is the 44.1k / 16 bit source)

For the first picture, in my analogy, it is similar to the output of NOS DAC
For the second picture, in my analogy, it is similar to the output of Modern OS DAC
For the thrid picture, in my analogy, it is similar to the output of upsampled music

Remember, they are all from the same source. I really like analogy... to me, it helps to explain to other people.

Let's think about cooking analogy,
the source (44.1k/16 bit) is the recipes
the DAC is the chef
the final music output is the dish

The master lab give you the recipes for a dish. Your hifi system "cook" it for you.
I would not say my music is better or worse than the one from the master lab as they are not comparable.
I won't compare a recipes with a final dish as I won't compare apple to orange.

ok, go back to the pictures above, let's use cooking analogy:

For the first picture, it's the final dish from restaruant NOS
For the second picture, it's the final dish from restaruant DAC
For the third picture, it's the final dish from restaruant UPSAMPLING

Again, let me remind you. They are all using the same recipes.

Someone prefer to look at the pictures like this:

View attachment 364768

Someone prefer to look at the pictures like this:
View attachment 364769

Again, same source, same pictures, just different presentations.

Bottom line, to me, I think it is all down to our own personal preference for what is "better music". Which restaurant you want to go to.

There is no right or wrong, correct or not, true or false, etc..

You have to remember again that all these images should always be viewed at the same size, and to make it directly comparable to audio at 44.1khz they should be viewed at something like 8K, and lets say a 32" display at a meter away, and when upsampling it to 16K viewed on the same display size and distance you will most likely not see any difference because your eyes can't resolve that extra fine details, just as your ears can't resolve that extra samplerate over 44.1khz.
Of course we might then still have issues such as with your DAC not being able to handle higher samplerates giving IMD that might trickle down in the audible range, but then you have faulty gear instead.

I'm not particularly convinced by it, but I wanted to mention another thing. One second of the original sound will always result in one second of reproduced sound, no matter what sampling rate you use in between. For images, 1 inch x 1 inch of the original picture will rarely result in the same size of reproduced picture and usually the higher sampling rate you use (i.e. more pixels), the bigger it will be.

No it's the exact same as in audio, that one second will always be one second regardless of the resolution and an image should always be viewed at the same size and distance no matter the resolution. If you scale up an image and compare it and the original at both 1:1 you are not doing a fair comparison.
This is one of the reasons why the myth of larger pixels on a camera sensor have less noise, because people have compared it 1:1 which give an unfair advantage to the lower resolution one.

 

[Tell]

They are poor analogies:
1. An amplifier and speakers can easily exceed the frequency of Red Book fs/2, but the best optical lenses are struggling to get close to or are worse (e.g. at the edge) than the pixel-per-mm/2 of a camera chip. This would be like CD comfortably getting to 20kHz but amplifiers and speakers all rolling off at 12kHz. This is the reverse model. You can NOT compare audio top end with fine edges in a picture!
2. Motion-picture technology limitations exist in the steps in vectors between separate samples. This has NO analogy in audio

No it's not poor analogies, because they are both digital systems complying under the same Nyquist-Shannon Theorem. As I explained earlier audio is a lot simpler with only time and amplitude, while images have two spatial dimensions, at least three channels in amplitude and when it comes to video a time dimension as well. But they both is bound by the same rules and in the end it's our senses that are the limiting factor. And fyi even the very old analog film of the moving train that the Lumiere brothers showed in 1986 have aliasing. It's of course experienced differently, but it's the exact same principle as in digital audio.

1. Yes audio have come way further than image technologywise, but that's why I do compare it to 8K and 16K since that's where I think our eyes no longer can resolve any differences even though the technology itself haven't come that far yet.
2. The exact same as in audio, except that, again, audio have come further technologywise.

 

[Tell]

Just want make sure that I fully understand what you mean.

Did you mean I cannot hear anything with frequencey higher than 22kHz so extra samplerate over 44khz won't help? or
Did you mean I cannot hear the difference in the audible range between the audio signal re-constructed using 44kHz and 768kHz?

Both.