I hear no difference when playing LP through my AV prepro whether digital bypass is enabled (all analog) or not (digitized at 96 kHz). Same is true when I run analogue signals through a digital EQ. When I rip a LP on my PC and play it back it sounds exactly like the LP. Therefore we can savely assume that the AD/DA process is transparent.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
What is the point of upsampling?
- Thread starter Tks
- Start date
As long as the DAC suppresses jitter artefacts below your audibility threshold Dual AES has no advantage.
Please watch the video -- it is well worth the effort. Nobody is arguing that you do not hear a difference, but that your assumptions about the "badness" in digital music are invalid. What we need are more folk who understand why their LPs may sound better and to push back on the recording industry to get it right.
Will do. Thanks. Appreciate the information you all provided.
OK, Don, I watched Monty's video. You guys are all engineers and math guys; I was an English/liberal arts major many, many years ago (so, I'm probably at the wrong Website ; -). So with that caveat, let me ask another question that I thought about after watching the video.
Say you have an original analog waveform: A. Then you have a digitally-sampled version of that original waveform: B. B has all those individual data points shown in the video. Then you add some mathematical wizardry (interpolation?) to join all those individual points into a perfect replica of the original analog waveform, which I will call C. I think that was what "Monty" was pointing to on his oscilloscope (?) to confirm that there was no data loss between A and B.
My question: does C contain exactly the same information as A?
No. Because of the finite precision of whatever number of bits you use, there will be a small "rounding error" (called "quantisation error"). That error results in a very low level noise (way lower than pretty much any source material) - below -95 dB for 16 bits, -140 dB for 24 bits.
This is not unlike the same process done using an analog path - you add some noise.
- Joined
- Feb 23, 2016
- Messages
- 20,793
- Likes
- 37,702
I don't want to complicate things too much. If you aren't used to it, please realize decibels are a logarithmic scale to make dealing with signals very different in size more manageable. A -95 db signal is .00178 % the size of the full scale signal. So quantization error is in that range. Using dither that can be reduced by a factor of 10 or more. So even with 16 bit the error vs perfect reconstruction could be in the order of 1 part per million or less. With 24 bit the error would be .0000064% or few parts per 100 million. Now none of our analog electronics can be that low in error as 24 bit for among other reasons thermal noise. With dither digital reconstruction does become very, very, very close to perfect with a bit of noise added.
As @Julf said they are not identical. There is resolution in time and amplitude. The ideal SNR due to amplitude quantization (conversion) is about 98 dB for a 16-bit converter. Dither (noise) is added to spread the samples around and that lets us resolve to below the quantization noise floor so it is really a little more than that. But that is still a lot compared to perhaps 60~65 dB for an LP. You are also sampling in time at 44.2 kS/s (or whatever) so the maximum frequency is limited to about 22 kHz without aliasing. Below that frequency, and above about 98 dB SNR, they will be essentially identical. There is no interpolation in the sense of adding information "between" the samples; filtering the high-frequency signal (above 22 kHz) removes (rounds) the edges of the stair steps and they disappear. A square wave ("stair-step") requires a string of high-frequency components to provide a fast edge -- see https://www.audiosciencereview.com/.../composition-of-a-square-wave-important.1921/ for an explanation. When you filter the DAC's output to eliminate the signal above 22 kHz (the Nyquist frequency, 1/2 the sampling frequency) the signal becomes nicely smooth and looks like any other analog signal (note DAC stands for digital-to-analog converter).
Bottom line is C will contain all the information for all practical purposes. Assuming a 16-bit CD-rate (44.2 kS/S) DAC, it will not contain frequency content above 22 kHz (ultrasonic) and will not quantize below about 98 dB (though we can hear to below that because of dither and the way our ears/brain filter out noise).
An analog signal path typically adds (much) more noise and distortion to the signal, and again look at the specs of your headphone or power amp and speakers for comparison to the DAC.
Here's another hand waving way to think about it: your LP provides SNR around 60 dB so the noise floor is about 1 part in 1000. The noise floor of a 16-bit DAC is about 1 part in 65,536. So yes, the digital signal is not the same, but the difference is 0.015 or 65.5 times (almost two orders of magnitude) smaller than the noise floor of the LP. That is, inaudible.
HTH - Don
Thanks, Don. I will have to read your comment, and those of the others who were nice enough to reply, a few times (at least) to make myself believe that I understand it -- at least, most of it.
At the very real risk of beating a dead horse, what accounts -- in your estimation -- for the audible difference that I hear between LPs and CDs? It's not my imagination; it's real and apparent. And, I also realize this is very, very subjective, but I love music and listen carefully to it -- it's not background stuff. I have a nice sound system that is highly-resolving musically, so equipment isn't really playing a big role in coloration.
I would appreciate your thoughts on this.
Thanks, Bob
Thanks for taking the time to reply. I will read it a couple of more times. I appreciate your comment -- as much as I understand of it ; -).
Bob
Thank you for taking the time to reply to my question. This is a learning experience for me.
Bob
I think I pointed out earlier on that there definitely is an audible and measurable difference between LPs and CDs. CDs are pretty much an audibly accurate and transparent reproduction of the source material, while LPs are not. LPs add a significant amount of distortion, noise, coloration and wow&flutter. Some people find the distortion and coloration (as well as low-frequency rumble and phase effects) pleasing (often because that is what they grew up with in their most formative years).
- Joined
- Feb 23, 2016
- Messages
- 20,793
- Likes
- 37,702
Also, LP's are going to have different mastering than CD's. You would expect some difference in sound from that.
Thanks.
Thanks. Makes me wonder what all is entailed in "mastering"?
Oh, lots of stuff. All depends on producer and mastering technicians. But for vinyl specifically, it means mixing low frequencies in mono, restricting amplitude (by using compression) to avoid needle jumping, and many other tricks.
I have not read the rest of thread but imagine everyone will say the same things, mainly:
- Mastering; and,
- Expectation bias.
Mastering is what many of us have found to be the difference. Mastering is one of the steps in the process that takes the sound at the recording microphone all the way to the CD or LP you play at home. Mastering is usually the last step and is when the sound engineer creates the "sonic balance" of the final recording. Effects like reverberation can be added, the equalization can be adjusted, and loudness adjusted (typically compressed). In short mastering greatly affects the sound you hear from the recording. A lot of CDs were remastered from the original recordings (tape or digital files) nd re-released in CD format; sometimes to preserve the songs, sometimes to "improve" the sound, and often both. The problem is that many CDs have dynamic range (difference between loudest and softest sounds) compressed and then the level boosted to make them louder. Louder generally sounds better to people, at least at first, and the record industry decided that making CDs louder would boost sales. Adjusting the equalization (EQ) to boost highs and/or lows was also sometimes done, and in some cases the mix (how the different voices and instruments are blended and leveled in the recording) were adjusted. Louder or not, the sound recorded to a CD is often quite different than that on an LP, and that is why they sound different to you (and to everyone else). Really nothing to do with the conversion to digital.
On the flip side, equalization (frequency response) is often adjusted on LPs to reduce the lows, and dynamic range may be compressed as well, to meet the limitations of the media. Another factor is that the higher crosstalk (left channel in the right and vice-versa) and distortion (often low-order) actually makes the sound "richer" and fuller" even though it is less accurate.
I have some CDs that are faithful to the original performance and they sound much better than the LP I have of the same recording. And many CDs that to my ears exhibit poorer performance after the remastering process (frankly, they suck). I know it is not the digital conversion process, and there are lots of others with the same complaint, including a bunch of folk who have analyzed the dynamic range and frequency response of their CDs to prove how they have been degraded.
@mitchco has a link to a recording a friend of his did showing what havoc loudness compression causes. A shame to take the huge dynamic range made possible by digital sound and squash it to much worse than the LP!
HTH - Don
Thanks, Don, again. I'm beginning to understand the process much better, and I appreciate why my original thoughts about digital sound were, as you diplomatically said, "invalid."
I don't think that "expectation bias" entered into my thoughts at all, because, frankly, when I spun that first vinyl record after many years of listening only to CDs, I wasn't expecting much. But, I was very pleasantly surprised. I liked the difference I heard very much, and I became curious.
Now, the question in my mind is why? I expect it might have to do -- not with my faulty rationale, before, of omitted information created by the digital sampling process -- but more with some of the distortion (that you folks have pointed out) inherent in vinyl. Perhaps that gives the sound a quality that I happen to enjoy, i.e., a less clinically-accurate one, but more akin to what I hear after the sound has bounced off the walls, people, etc., in a concert hall.
I have found bad-sounding LPs and CDs, so the reproduction of the music by careless humans is a major factor, too.
And, all this started with me wanting to learn a little about "upsampling." Wow. Be careful what you wish for. ; -)
Thanks for your patience and taking the time to educate (a little) a non-engineer. I've enjoyed it.
Bob
@SIY in another thread pointed to an AES paper and article in Linear Audio by Michael Urwins on his investigation on the reasons behind the 'vinyl attraction'.
https://www.audiosciencereview.com/...nderstand-the-appeal.6430/page-23#post-170710
Here is a link to Urwins' Linear Audio article which included more details than the e-brief. It is a pretty long read at 26 pages, but if you are interested in this topic, it is well worth your time.
https://linearaudio.nl/sites/linearaudio.net/files/v10 mu.pdf
Last edited:
Similar threads
