What is the point of upsampling?

[Julf]

I don't agree with you, but hearing is a very subjective thing.

Hearing and preference are subjective things. Accurate and transparent reproduction of a signal isn't.

In my case, I have been surprised by the audible difference between CDs and LPs.

We all agree there is a difference. The vinyl record process is not transparent or very accurate - it is saddles with distortion, noise and coloration.

I don't hear a greater dynamic range or deeper bass with digital, in fact, it's just the opposite.

You of course hear what you hear, but thousands of measurements seem to disagree with you.

My guess is that the digital sampling process, of necessity, leaves some music UNsampled. This may be relatively small, but I'm beginning to think that it is very important to imparting realism and presence to the music. Again, my opinion only.

Only a guess, not supported by evidence. It has been proven, many, many times, that if you record the output of a LP player digitally and then reproduce the digital recording, it sounds exactly the same as the original LP player output. Doesn't work the other way around.

Also keep in mind that pretty much all mainstream vinyl cut since the mid-1980s has been converted to digital and back in the cutting lathe.

 

[Julf]

Just now a new example: Various - Once Upon A Time In Hollywood (Original Motion Picture Soundtrack)
https://www.audiosciencereview.com/...-are-we-listening-to-right-now.40/post-254572

Vinyl: DR11

CD: DR9

I hope you are aware that the DR numbers calculated by that somewhat simplistic algorithm aren't comparable.

 

[bwinlr]

So all digitally recorded music (which is the vast majority) has lost vital information yet when such recordings are released on vinyl it magically returns ?

No, not at all. Again, in my opinion, it depends entirely on the original source material. The LPs I"ve been buying were all originally recorded on analog tape. It would make no sense to buy a record of a performance that was digitally recorded, but you find a bunch of those out there today.

Vinyl records can't restore music that was never captured originally. My point is that the digital process of "sampling" the wave form at different points leaves out music between the sampling points. That's simply part of the process, and is undeniable. As I understand it, the "interpolation" process is designed to smooth out the sampling "steps" by inserting sound that really wasn't there originally to make the waveform smoother.

 

[bwinlr]

I also "got back into" records about 6-7 years ago. I have a good deal of records (around 1400 LPs) and a good deal of CDs (around 4000+). IME when people jump back into vinyl they are in the honeymoon period and the downsides haven't really set in yet. It's also easy to knock CDs if you just bought whatever was in front of you with without researching what to buy first. With CDs, as with LPs, everything needs to be taken on a case by case basis. There are many well mastered CDs out there, and many poorly mastered ones, and yes newer remasters are often horrible (though not always). Same deal with records - lots of crappy vinyl out there new and old mastered poorly or pressed poorly or that used an inferior source. Now, I enjoy both formats (and lossless digital files as well) but have a realistic view of the technical shortcomings of vinyl.

I do, too. I'm not suggesting that all CDs are bad -- I have many more of them than vinyl records, and enjoy them. And, you may be right about the "honeymoon" effect, but I was surprised by how much I liked the first vinyl records I spun after years of having dismissed the format. There was an audible -- and very pleasurable -- difference in presence, immediacy and "warmth" to the vinyl.

 

[Julf]

As I understand it, the "interpolation" process is designed to smooth out the sampling "steps" by inserting sound that really wasn't there originally to make the waveform smoother.

No, not at all. Are you familiar with the Shannon-Nyquist sampling theorem?

 

[bwinlr]

Or personal and mind biases. Preference does not necessarily mean better in a profound sense.

That certainly could be operating with me, but it would be impossible to determine. And, I agree with you about the difference between "preference" and "better."

 

[DonH56]

My point is that the digital process of "sampling" the wave form at different points leaves out music between the sampling points. That's simply part of the process, and is undeniable. As I understand it, the "interpolation" process is designed to smooth out the sampling "steps" by inserting sound that really wasn't there originally to make the waveform smoother.

The information that is "thrown out" is above the anti-alias/image filter at 20 kHz and up. Nothing is "thrown out" below that frequency, and amplitude information is not lost. Furthermore time resolution is not lost; it is set by the aperture time and not the sampling rate, so is often in the ns or ps range (way, way above the audio band). Those stair steps fearmongers love to show are not present in the signal after filtering. Please check out the video posted earlier in this thread to see how it works. There is a lot of misinformation spread about digital audio.

 

[Blumlein 88]

Try this video. The one Don is referring to.

Keep in mind Monty is using all high quality analog monitoring gear to show you what digital is doing to the audio.

 

[Leporello]

Vinyl records can't restore music that was never captured originally. My point is that the digital process of "sampling" the wave form at different points leaves out music between the sampling points. That's simply part of the process, and is undeniable. As I understand it, the "interpolation" process is designed to smooth out the sampling "steps" by inserting sound that really wasn't there originally to make the waveform smoother.

Intuitively, all this may sound very convincing and 'undeniable' to you - and to many others. But digital audio cannot be grasped intuitively.

You prefer the sound of vinyl to the sound of cd. That is perfectly ok. But you then try to find a technical justification for your subjective preference by making claims that are not subjective at all. What is more problematic is that they - to the best of our current knowledge - are not true.

 

[bwinlr]

No, not at all. Are you familiar with the Shannon-Nyquist sampling theorem?

No, not at all. Are you familiar with the Shannon-Nyquist sampling theorem?

Again, new to this reply process, hence the "oversampling." ; -)l

No, I'm not familiar with the Shannon-Nyquist sampling theorem, and I appreciate you mentioning that. I will look at it. Thanks.

 

[Julf]

No, I'm not familiar with the Shannon-Nyquist sampling theorem, and I appreciate you mentioning that. I will look at it. Thanks.

As you do, remember that it is not a theory, but a theorem (just like the Pythagorean theorem). Something that has been formally proven based on mathematics.

What it says is basically that you can reconstruct a sampled, bandwidth-limited signal perfectly. You are not "approximating" or "inserting stuff that wasn't there". You are reconstructing the original signal.

 

[bwinlr]

Intuitively, all this may sound very convincing and 'undeniable' to you - and to many others. But digital audio cannot be grasped intuitively.

You prefer the sound of vinyl to the sound of cd. That is perfectly ok. But you then try to find a technical justification for your subjective preference by making claims that are not subjective at all. What is more problematic is that they - to the best of our current knowledge - are not true.

A bit harsh, but probably true in some respects. I really wasn't trying to "justify" my subjective experience, it was simply a "guess" as to what might account for it. I enjoy CDs and LPs, but I do hear an audible difference.

I think much of this discussion inevitably goes back to the old argument between sound that "measures well" versus music that "sounds good." I'm very much in the latter camp. I read John Atkinson's measurements in "Stereophile" magazine, and note how the reviewer's opinion of the piece of equipment tested sometimes varies from his measurements. It's interesting.

 

[BDWoody]

I read John Atkinson's measurements in "Stereophile" magazine, and note how the reviewer's opinion of the piece of equipment tested sometimes varies from his measurements. It's interesting.

It would be a lot more interesting if the listener implemented any controls on the listening process. Uncontrolled subjective 'reviews' are basically marketing copy, and almost always directly correlate to $$$.

 

[Julf]

I think much of this discussion inevitably goes back to the old argument between sound that "measures well" versus music that "sounds good." I'm very much in the latter camp. I read John Atkinson's measurements in "Stereophile" magazine, and note how the reviewer's opinion of the piece of equipment tested sometimes varies from his measurements. It's interesting.

To me that is a bit like wearing soft focus eyeglasses to a photo exhibition just because you like soft focus on portrait pictures (or rose-colored glasses to a Rembrandt exhibition because you like the colors a bit warmer), but that is just my personal view.

 

[Jaysz]

Isn't upsampled what lots of the hires downloads are.

 

[bwinlr]

To me that is a bit like wearing soft focus eyeglasses to a photo exhibition just because you like soft focus on portrait pictures (or rose-colored glasses to a Rembrandt exhibition because you like the colors a bit warmer), but that is just my personal view.

And, I respect your right to hold that view, but I simply don't agree with it. But, I realize that this site is titled "AudioSCIENCEreview," and that would explain many of the preferences expressed.

 

[Julf]

And, I respect your right to hold that view, but I simply don't agree with it. But, I realize that this site is titled "AudioSCIENCEreview," and that would explain many of the preferences expressed.

Yes, many of us have actually studied and worked in the field of audio, and understand the science and engineering.

 

[bwinlr]

It would be a lot more interesting if the listener implemented any controls on the listening process. Uncontrolled subjective 'reviews' are basically marketing copy, and almost always directly correlate to $$$.

In certain publications, I have a feeling you're absolutely right about the reviews being "marketing copy." However, I think some other publications have more journalistic integrity.

 

[Julf]

Isn't upsampled what lots of the hires downloads are.

Yes. Doesn't make them any better (just like pouring a pint of beer in a gallon container doesn't make it any better.

 

[Julf]

In certain publications, I have a feeling you're absolutely right about the reviews being "marketing copy." However, I think some other publications have more journalistic integrity.

Yes, mostly professional publications, but even those are mostly paid by advertising. As to "hi-fi" and "audiophile" magazines, I think any pretenses of journalistic integrity went out the window 30-40 years ago.