Skip to main content

Topic: Vinyl is equivalent to which digital bit-depth and sampling rate? (Read 63105 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #50
Maybe this is the right discussion to ask a question that came into my mind after watching Monty's video.

I understand that only one band limited signal is able to pass the samples perfectly to reconstruct the original band limited signal.
But how does the DAC know how to construct the signal between two samples. Does it do some math inside? Or to be more concrete: how are all the volts created that form this smooth curve?

  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #51
It's convolved with (filtered by) a sinc pulse. That's it. Perfect reconstruction. It's pretty easy to do in practice, with ridiculous (i.e. far more than you need) amounts of precision.


However, you can use anything or nothing for the reconstruction filter. All the differences between what you have and what you want are above fs/2 (i.e. above 22kHz for a CD), hence your ears can't hear them, hence they don't matter. Really. Let your ears be the reconstruction filter.

Unfortunately ultrasonic junk like that interferes with non-ideal real-world equipment, so it's good practice to filter it out. Otherwise intermodulation distortion can cause the ultrasonic junk to have an effect within the audible range, or even damage equipment.


Also, the filter transition (between pass band and stop band) is often softened - so that instead of letting through everything up to 22.05kHz and stopping everything over 22.05kHz, it lets through 21kHz, stops 22kHz, and has a transition between. Or even it lets through everything up to 18kHz, stops everything above 22kHz, but has a 4kHz transition band in between where things are reduced but not eliminated. This is totally equivalent to using a perfect sinc pulse, and then applying the softened filter - so it's still conceptually "perfect" but then it also reduces the amplitude of everything above 18kHz.

Cheers,
David.

  • mjb2006
  • [*][*][*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #52
What needs to be demonstrated is that the filter removing the frequencies above the Nyquist removes the between-sample squiggles such that the samples do represent this smoothed-out curve perfectly, and that this smoothing doesn't result in any crucial information being lost. Thus, those squiggles weren't contributing anything to the frequency components below the Nyquist. (That's right, right?)

IIRC, Monty's video shows some of this, but relies too much on his calm reassurances rather than a demo of every step of the process, both in simple terms and with a real-world example.
Monty's video includes exactly this from about 4 minutes in...
http://xiph.org/video/vid2.shtml
...showing exactly what you want from about 5 minutes with high frequency sine waves. Square waves at 17:50.

Well, just to be sure, I watched them again, and my impression is the same. The fundamental concepts people need are provided by the videos in bits and pieces, a little at a time. You get a little bit in the first video at 4:30, and then in the second video at 7:00, 17:50, and 21:00. And then there's even some in Monty's responses to comments on the evolver.fm repost of the 24/192 article.

It's also all in different contexts. The stuff at 7:00 is great but it's all in the context of stairsteps. The stuff at 17:50 gives you some important info, but it's in the context of taking the mystery out of the Gibbs Effect. There's nothing demonstrating music and analog sources in there... and there shouldn't be, at first, but people need to see both the simplified fundamental concept and at least one example they can relate to.

It's just a lot to ask people to set aside whatever they believe about analog v. digital, watch all this video, glean very specific pieces of info from different parts of it, and integrate it with their impression of what's going on in record grooves and the digital sampling thereof. Pointing to Monty's work is the best we can do right now, but we should be able to do better.
  • Last Edit: 04 September, 2014, 10:38:38 AM by mjb2006

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #53
Speaking as a self-selected sample of one...

Monty's videos were the turning point in my getting a better understanding of digital audio. Yes, I thought that sampling meant that stuff was left out, and that the wave form of digital music actually looked like steps --- and the more the better.

I'm a maths duffer, and graphs tend to freeze my brain. Even so, after watching Monty, I've been able to take in stuff from others who have the knack of making the depth of this stuff accessible to the likes of me, like JJ Johnston. I still have not reached the point where I can take in the equations, or where I do much more than take all that sync stuff except on trust from real mathematicians.

It may not be true of sample rates, but the more the merrier is certainly true of down-to-earth and accessible explanation of this stuff. Whoever wants to pile in with more will find a willing audience of people like me.

The other basic misconception amongst audiophiles, is that they suppose it to be a sampling theory, rather than a sampling theorem. Even the academic ones don't seem to appreciate the big difference.

EDIT:Whoops, can't spell theorem. Just shows how academic I'm not!
  • Last Edit: 04 September, 2014, 01:46:35 PM by Thad E Ginathom
The most important audio cables are the ones in the brain

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #54
However, you can use anything or nothing for the reconstruction filter. All the differences between what you have and what you want are above fs/2 (i.e. above 22kHz for a CD), hence your ears can't hear them, hence they don't matter. Really. Let your ears be the reconstruction filter.


OK, I'm risking showing my ignorance here, but as I understand it, if we don't have the correct filter, the frequency response trails off gradually well before we approach Nyquist. Intuitively, that is because we are simply joining the dots with a 'non-ringing' filter which cannot create the peaks between samples that the sinc filter does.

Quote
It is already at -3.2dB by 20kHz

http://myweb.tiscali.co.uk/g8hqp/audio/CDsample.html

edit: quote & link, plus added "well before we approach Nyquist"
  • Last Edit: 04 September, 2014, 03:15:14 PM by Green Marker

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #55
However, you can use anything or nothing for the reconstruction filter. All the differences between what you have and what you want are above fs/2 (i.e. above 22kHz for a CD), hence your ears can't hear them, hence they don't matter. Really. Let your ears be the reconstruction filter.


OK, I'm risking showing my ignorance here, but as I understand it, if we don't have the correct filter, the frequency response trails off gradually well before we approach Nyquist. Intuitively, that is because we are simply joining the dots with a 'non-ringing' filter which cannot create the peaks between samples that the sinc filter does.

Quote
It is already at -3.2dB by 20kHz

http://myweb.tiscali.co.uk/g8hqp/audio/CDsample.html

edit: quote & link, plus added "well before we approach Nyquist"


Not the best article in the world.

Producing ADC/DAC pairs that have loss on the order of -0.1 dB at 20 KHz was commonly done in the 1980s.  The Sony CDP 101 was one of the two first CD players introduced to the public in 1982, and it had analog reconstruction filters. The Philips CD-100 was introduced about the same time and it had digital filters and oversampling.

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #56
Producing ADC/DAC pairs that have loss on the order of -0.1 dB at 20 KHz was commonly done in the 1980s.  The Sony CDP 101 was one of the two first CD players introduced to the public in 1982, and it had analog reconstruction filters. The Philips CD-100 was introduced about the same time and it had digital filters and oversampling.

True I'm sure, but not answering the point I was i.e. that any, or no filter will do.

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #57
Producing ADC/DAC pairs that have loss on the order of -0.1 dB at 20 KHz was commonly done in the 1980s.  The Sony CDP 101 was one of the two first CD players introduced to the public in 1982, and it had analog reconstruction filters. The Philips CD-100 was introduced about the same time and it had digital filters and oversampling.

True I'm sure, but not answering the point I was i.e. that any, or no filter will do.



That of course depends on "will do" means. If we are talking about LPs or analog tape, then - 3.2 dB @ 20 KHz at FS is pretty darn good.

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #58
That of course depends on "will do" means. If we are talking about LPs or analog tape, then - 3.2 dB @ 20 KHz at FS is pretty darn good.


So in our hypothetical conversation with the vinyl-oriented audiophile we have gone from digital audio being theoretically exact, to saying that joining the dots with straight lines (just as he imagined they were) and being -3dB down at 20kHz is darn good. Hmm...
  • Last Edit: 04 September, 2014, 06:05:28 PM by Green Marker

  • saratoga
  • [*][*][*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #59
However, you can use anything or nothing for the reconstruction filter. All the differences between what you have and what you want are above fs/2 (i.e. above 22kHz for a CD), hence your ears can't hear them, hence they don't matter. Really. Let your ears be the reconstruction filter.


OK, I'm risking showing my ignorance here, but as I understand it, if we don't have the correct filter, the frequency response trails off gradually well before we approach Nyquist.


If you have no filter, the frequency response is infinite, and therefore never trails off.  You'll accurately reproduce many images of the signal at higher and higher frequencies.

Of course an infinite bandwidth is impossible in the real world, so even if you don't include a filter, something in your circuit will act as one, and eventually the frequency response rolls off.  In some applications the ability to produce or record image/aliased frequencies is desired, in which case no filter is used.  In this case, the response typically rolls off quite badly by about 2 or 3 times the Nyquist limit, although that can be tuned and in some cases is much greater.

Intuitively, that is because we are simply joining the dots with a 'non-ringing' filter which cannot create the peaks between samples that the sinc filter does.


If the bandwidth is infinite you don't join the dots at all.  The signal instantly reaches the value specified, and then instantly decreases back to zero again until the next sample comes along.  This is what enables reproduction of image frequencies (replicas of your signal at frequencies above Nyquist).  This is not physically possible of course, so in practice you'll generate little sinc(x) functions (or something very close to them) around each sample.  As the bandwidth increases the sinc() functions get narrower and more delta-like.

  • Kohlrabi
  • [*][*][*][*][*]
  • Global Moderator
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #60
So in our hypothetical conversation with the vinyl-oriented audiophile we have gone from digital audio being theoretically exact, to saying that joining the dots with straight lines (just as he imagined they were)
Who said that? It doesn't matter how you join the dots, it's just a graphical representation. Yes it's a stairstep, or straight lines, or wavy spline lines. But this is completely, utterly unimportant, and just a pretty picture for visualization. Sampling works the same regardless of how your software or scope represents the joining of sample points.
  • Last Edit: 04 September, 2014, 06:30:09 PM by Kohlrabi
It's only audiophile if it's inconvenient.

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #61
So in our hypothetical conversation with the vinyl-oriented audiophile we have gone from digital audio being theoretically exact, to saying that joining the dots with straight lines (just as he imagined they were)
Who said that? It doesn't matter how you join the dots, it's just a graphical representation. Yes it's a stairstep, or straight lines, or wavy spline lines. But this is completely, utterly unimportant, and just a pretty picture for visualization. Sampling works the same regardless of how your software or scope represents the joining of sample points.

Ah, you see, this is exactly the problem. We're not talking about a mere representation, but literally how the dots are joined together before feeding the signal into the amp. We are now being told that any filter at all will do, or even no filter at all, which is not what the Xiph video says, for one.

To say that a filter-less DAC does not have a roll-off before Nyquist because in that case the frequency response is "infinite" is the very definition of semantics. Using a theoretically-correct filter (or as close as is practical) will give us a perfect (or as near as practical) response. By leaving the filtering to the amp or the listener's ear will give us a perceptible treble roll-off (and other nasties, probably). Why can't we simply agree that a digital sampling system needs a theoretically-correct (sigh, or as near as practical) reconstruction filter?

  • saratoga
  • [*][*][*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #62
To say that a filter-less DAC does not have a roll-off before Nyquist because in that case the frequency response is "infinite" is the very definition of semantics.


In so much as it is a trivially true statement, yes.

By leaving the filtering to the amp or the listener's ear will give us a perceptible treble roll-off (and other nasties, probably).


No absolutely not.  Otherway around:  you can't have roll off without something to provide filtering.  An infinite bandwidth means an infinitely high frequency response and zero roll off.

Does that make sense?  Its only once you have something filtering (either a reconstruction filter or else the limited bandwidth of the wires and chips themselves) out higher frequencies that you get roll off (which makes sense since roll off is literally just the filtering out of higher frequencies.

Why can't we simply agree that a digital sampling system needs a theoretically-correct (sigh, or as near as practical) reconstruction filter?


It depends on what you mean by "needs".  In practice it probably doesn't matter if the filter cuts off near Nyquist or not, although its a good idea because it will tend to make things a lot easier for your speakers afterwards.  If you're going to pump a lot of ultrasound into your equipment, you'll need to make sure they have the capacity to handle that power, and low enough distortion that you don't hear any intermodulation.

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #63
No absolutely not.  Otherway around:  you can't have roll off without something to provide filtering.  An infinite bandwidth means an infinitely high frequency response and zero roll off.

Does that make sense?  Its only once you have something filtering (either a reconstruction filter or else the limited bandwidth of the wires and chips themselves) out higher frequencies that you get roll off (which makes sense since roll off is literally just the filtering out of higher frequencies.


Here's a link to a document that contains a picture of the frequency response of an unfiltered DAC. It droops all the way to Nyquist and you get images out to infinity...

http://www.analog.com/static/imported-file...ials/MT-017.pdf

The energy in the images are useless to us and cannot be heard (they're ultrasonic) so "perceptibly" (as I said) there is a treble roll-off without the correct filter. An inverse sinc filter (or as near as practical, as I have to state round here, every time!) will correct the roll-off and suppress the images. Perfect!

edit: *inverse* sinc filter
  • Last Edit: 04 September, 2014, 07:47:30 PM by Green Marker

  • saratoga
  • [*][*][*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #64
No absolutely not.  Otherway around:  you can't have roll off without something to provide filtering.  An infinite bandwidth means an infinitely high frequency response and zero roll off.

Does that make sense?  Its only once you have something filtering (either a reconstruction filter or else the limited bandwidth of the wires and chips themselves) out higher frequencies that you get roll off (which makes sense since roll off is literally just the filtering out of higher frequencies.


Here's a link to a document that contains a picture of the frequency response of an unfiltered DAC.


That is not what you have linked.  That is actually the frequency response of a zero order hold DAC.

You may find this wikipedia article helpful in understanding that document:

http://en.wikipedia.org/wiki/Zero-order_hold

It droops all the way to Nyquist and you get images out to infinity...


A zero order hold DAC does not have an infinite bandwidth.  As your link notes, it has a 3.92 dB loss at Nyquist, meaning that its bandwidth is less than half the sampling rate.  In contrast, a typical DAC will have a bandwidth greater than half the sampling rate, and an infinite bandwidth DAC actually has an infinite bandwidth.

An inverse sinc filter (or as near as practical, as I have to state round here, every time!) will correct the roll-off and suppress the images. Perfect!


This is true for a zero order hold DAC, but not for a real DAC, which is what people in this thread are discussing...

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #65
That of course depends on "will do" means. If we are talking about LPs or analog tape, then - 3.2 dB @ 20 KHz at FS is pretty darn good.


So in our hypothetical conversation with the vinyl-oriented audiophile we have gone from digital audio being theoretically exact, to saying that joining the dots with straight lines (just as he imagined they were) and being -3dB down at 20kHz is darn good. Hmm...


No, the  hypothetical conversation with the vinyl-oriented audiophile got derailed from a middle-of-the-road well-designed  modern digital audio system to some Frankenstein system that someone made up.

There actually were a few cheapo-cheapo DACs with no output filters in the middle 1990s, but they are long gone. And there a few naive audiophiles who have built retro-freako, one-of-a-kind basement project DACs with the same problems. You can't stop people from doing crazy things in the privacy of their own homes.

Let's clarify this. The digital domain is as both theoretically and practically exact unless someone goes out of their way to introduce changes.

Unfortunately the analog systems on either side of the digital domain are as crappy and inexact as they ever were for a given amount of processing.  The big improvement over the past 20 or more years is that by moving most of the processing and all of the storage and data transmission into the truly perfect digital domain, we've done away with the need for most of that analog crap.
  • Last Edit: 04 September, 2014, 08:32:01 PM by Arnold B. Krueger

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #66
No, the  hypothetical conversation with the vinyl-oriented audiophile got derailed from a middle-of-the-road well-designed  modern digital audio system to some Frankenstein system that someone made up.


An excellent way of putting it!

  • 2Bdecided
  • [*][*][*][*][*]
  • Developer
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #67
To say that a filter-less DAC does not have a roll-off before Nyquist because in that case the frequency response is "infinite" is the very definition of semantics. Using a theoretically-correct filter (or as close as is practical) will give us a perfect (or as near as practical) response. By leaving the filtering to the amp or the listener's ear will give us a perceptible treble roll-off (and other nasties, probably). Why can't we simply agree that a digital sampling system needs a theoretically-correct (sigh, or as near as practical) reconstruction filter?

I refer you to my previous posts.
This is perfection:
http://www.hydrogenaud.io/forums/index.php...st&p=873787

The second half of this is still perfection within the audible range:
http://www.hydrogenaud.io/forums/index.php...st&p=873795


Let me explain it further. The first filter (anti-alias filtering, before sampling) is vital. If it's wrong, and lets in any frequency components above fs/2, the audio signal within the audible range is irreversibly damaged.

If you want to see the same waveform at the output, the second filter must be perfect too. Then you can visually (and mathematically) demonstrate that the overall process is perfect, and sampling does nothing to the signal.

If you don't filter, and output instantaneous (or near instantaneous) samples, then what you get is still perfect within the audible range. All the differences are above 22kHz.

If you don't properly "filter", but just output the sample values with zero-hold (i.e. a stair-step), then the frequency response gets multiplied by a sinc filter*: 3dB down at fs/2. But this is a really important thought experiment: if you still don't do "proper" filtering, but take the stair-steps, add a little filter that's the inverse of the sinc filter within the audible range (i.e. something that's does nothing at DC, and boosts 3dB at fs/2, with the right slope in between) the result, within the audible range, is perfect. The waveform looks really strange - steppy and now with little flicks/overshoots on the edge of each step - yet within the audible range, the result is perfect. All the visual difference is due to differences outside the audible range. The wierdy steppy waveform with flicks is exactly what you want in the audible range, plus lots of ultrasonic junk that you can't hear.

Get it? Perfect reconstruction is easily possible. It's desirable because it lets you look at the waveform as say "it's exactly the same", and it present a clean and easy signal to amplifiers and speakers, free from ultrasonic junk that may make them distort. But if you do no reconstruction at all, or stair-step reconstruction plus a little high frequency boost, the signal within the audible range is STILL perfect. All the stair-steppy-ness, and even the weird beating / amplitude modulation you see when sampling, say, 20kHz at 44.1kHz - all the stuff you see that wasn't in the original signal is above 22kHz, and so (if speakers didn't have a tendancy to distort when fed with high levels of ultrasonics, making the inaudible become audio) would be inaudible.


The point of all this? Some people like to say "ah well, digital sampling might be perfect in theory, but you can't make the filters good enough for it to be even half-decent in practice". Rubbish. 1) you can. 2) you don't even need to (except your speakers would prefer it if you tried to do a reasonable job!).

Cheers,
David.

P.S. If the filters are imperfect within the audible range, then they change the signal in the way that they change the signal. No more, no less. It's not magic. So if someone builds a filter that attenuates the signal by 3dB at 18kHz, then that's what it does. But it doesn't do something worse than that just because it's part of a DAC. The worst anyone ever did ina  DAC filter is still better than 99% of cartridges used for playing vinyl, and despite audiophile folklore, it doesn't magically become worse just because it's applied to digital audio rather than analogue.


* = a sinc shape in the frequency domain, not the time domain - don't get confused by sinc cropping up again here. You may recall from a previous post that you want a sinc filter - but that was a sinc shape in the time domain which gives you a flat response in-band in the frequency domain. Whereas here you have a sinc shape in the frequency domain, because of all those unwanted flat-tops in the waveform in the time domain. The time and frequency domains are related like this - it's a fascinating subject, but probably not one for today.

EDIT: spelling!
  • Last Edit: 05 September, 2014, 12:55:07 PM by 2Bdecided

Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #68
No, the  hypothetical conversation with the vinyl-oriented audiophile got derailed from a middle-of-the-road well-designed  modern digital audio system to some Frankenstein system that someone made up.


An excellent way of putting it!


Nevertheless, the devil must be given his due. Even doing something crazy by leaving off the final low pass filter produces a system that is still far better than vinyl.

  • xnor
  • [*][*][*][*][*]
  • Developer
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #69
A sinc filter is an ideal low pass filter.
DC to fc: 1 = 0 dB
fc: 0.5 = -6.02 dB
fc up: 0 = -inf dB
response (pi on the x-axis would be fc)

Zero-order hold on the other hand is simply a bad low pass filter, like the rectangular window.
Roll-off to fc (about -4 dB), nulls only at multiples of the original sample rate.
response (0.5 on the x-axis would be fc)
  • Last Edit: 05 September, 2014, 07:59:35 AM by xnor
"I hear it when I see it."

  • cdroid
  • [*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #70
I've read some of your responses and I'm slightly upset to be perfectly honest. First allow me to respond by qualifying myself.
No, I am not a hobbyist. I am a student with a certificate in digital audio and I'm currently working towards an AA. I spent all morning discussing digital audio theory with my professor who would adamantly disagree with the responses I've read. No, my post was not riddled with audiophile lore, it was riddled with facts from my course books. I have also chosen this as my career path and am literally paid to know these things.

First I will discuss the MP3 and DAC related posts that many people commented on.
Yes, MP3 lossy compression actually is a serious problem and we can actually test that very simply;
Grab a CD with Red Book specifications. Now, set your iTunes to decode to AAC. Burn the disk to AAC. Now listen. Then set your iTunes to decode to WAV. Now listen.
I did this test a couple days ago because I'm looking at upscaling my entire iTunes library when I sync to a NAS server on my internal network, and I noticed a huge improvement even on small shotty speakers over my shotty DAC on my iMac.
If you can't hear the difference from this test, then you're probably a hobbyist with untrained ears. Look at the file difference. The WAV files are going to be almost 5 times the size of AAC files. The AAC encoder is dropping a whole lot more audio information than you'd care to admit. And put AAC vs. WAV aside for a second, I think we can all agree that AAC is better quality than MP3. How much more will the difference between MP3 and WAV be?

Now let's start talking DACs. Just so you have a feel of where I'm coming from; I no longer am impressed by Avid Venue DACs (or preamps, but that's another issue) or the last couple generations of Pro Tools HD DACs (again, Avid Preamps suck). I was, however, impressed with the MIDAS and Digico DACs. In the studio I would be happy with UAD Apollo DAC as well as Apogee DAC.
In the more affordable range, I used to use PreSonus for DAC and I thought it was hot but that was about 5~6 years ago now. Now I use the new Behringer DACs and I'm very happy for the price. When I need high fidelity quality for mixing or mastering, I use my MOTU interface and have been satisfied.
Obviously I don't expect the consumer market to invest in pro audio equipment just to listen to music (you should), so I would recommend the Behringer UCA222; it will do the trick. I haven't personally tried the consumer grade products or USB headphone amps but I'm sure they'd be sufficient.
I also just realized that perhaps I should qualify what I meant when I said that computer DACs don't sound good. If you own a tower and you have anything near a half decent sound card in there, then you've likely got a quality DAC. What I was talking about is laptops or all in one computers with the sound card built into the motherboard. Laptops, phones, and All in one computers have always packed 10 pounds of crap into a 5 pound bag (this is why I'm going to build a tower to replace my Macbook Pro) and you can't trust those DACs to do their job. I was using my MacBook Pro to listen to music, and then substituted my UCA222 I mentioned earlier for my built in DAC and noticed a pretty massive quality increase.

I saw a person argue that DACs have been stable for 20~25 years (I do not kid you, go read the second or third post from mine, it will give you a good laugh if you understand digital audio and computer theory) which is preposterous since digital recording was only available to the consumer about 25 years ago. My professor who has used Pro Tools since it was called Sound Tools would have argued that even 20 years ago it was new and unstable technology, and if you asked him I think he'd argue that even the last generation of Pro Tools HD DACs are unstable (the new architecture of HDX is a lot better).
There are a few things that effect DAC quality (besides being inherently vulnerable to errors) including noise, harmonic distortion, max sample rates, and dynamic range.

One of the biggest issues with cheaper DACs like the ones in phones and the like is the signal to noise ration. When you use cheap parts you're going to get noise.

I don't feel like reading the rest of the responses tonight because I really am just not in the mood for that. Surprise; it upsets me to spend 8 hours with a college professor and another hour or two reading suggested course material only to log onto my computer to have a hobbyist refer me to some youtube videos and wiki pages.
To the smart alek referring me to wiki pages; read Audio Engineering 101. It will serve as a decent intro so you can launch into some better primers to the audio world like Principles of Digital Audio: Sixth Edition. The first chapter of Principles of Digital Audio will go over computer language such as binary and hex and help you understand exactly what lossy encoders do.

  • cdroid
  • [*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #71
I forgot to reply to the post that irritated me the most. I had said that DSP (digital signal processing) has not surpassed analog processors, and some dude who admitted he has no experience said he doubts it. For starters, when you see Chris Lord Algae selling his LA-2A, then you can believe that the CLA-2A plugin has surpassed analog. In the meantime, all the pros are exporting their Pro Tools HDX recordings back into an analog soundboard using analog gear for inserts and there is a reason for that.

Digital coding has become a lot better, especially since the UAD 2 procs are using the new SHAARC chips. Those UAD plugins have become good enough that The Black Keys are mixing all in the box, but they're pioneering that trend and no one else (except post engineers) are really following suit. Even though the new Pultec eq from UAD is a lot better than the legacy one, it doesn't compare to the classic Pultec tubes (or the new ones since Pulse Technics is back on Sweetwater). People are still buying or building tube LA-2As even though a UAD proc with the LA-2A plugin collection is cheaper. This is because mathematics can't replace the sound of a good tube. The plugins are just algorithms, and these algorithms become increasingly complex the closer these engineers try to model the way the classic gear responds to different signals and setting tweaks.
Digital plugins that aren't modeled after analog gear are missing most of the harmonic distortions that are sonically pleasing. You won't see engineers trading out their racks of gear for UAD, and you won't see engineers trading out their analog soundboard for Slate VCC. Anyone who owns a Studer would laugh at the A800 plugin for UAD.

Now, go talk to guitarists about Amp modeling. Go ahead, I dare you to open that can of worms. Rush is the only major band using amp modeling, and they're still using real heads.

  • splice
  • [*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #72
Sigh... this place needs a :facepalm: icon.

I've read some of your responses and I'm slightly upset to be perfectly honest. First allow me to respond by qualifying myself.
No, I am not a hobbyist. I am a student with a certificate in digital audio and I'm currently working towards an AA. I spent all morning discussing digital audio theory with my professor who would adamantly disagree with the responses I've read. No, my post was not riddled with audiophile lore, it was riddled with facts from my course books. I have also chosen this as my career path and am literally paid to know these things. ...


You might take a moment to examine the CVs of the people in this discussion. You will find they used to be like you once, but now they're older and know a lot more.


First I will discuss the MP3 and DAC related posts that many people commented on. ...


If you did your homework, you would know that some of the people here code perceptual compression algorithms and design DACs for a living. The phrase "teaching your grandmother to suck eggs" springs to mind. 

... some youtube videos and wiki pages.  ...


Do you know who Chris (Monty) Montgomery is and what he does for a living?
Regards,
   Don Hills
"People hear what they see." - Doris Day

  • splice
  • [*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #73
... when you see Chris Lord Algae selling his LA-2A ...


If you're going to hold Chris up as an example, you ought to at least spell his name correctly.

I think the rest of your post would be better posted over in the Purple Place.
Regards,
   Don Hills
"People hear what they see." - Doris Day

  • sld
  • [*][*][*][*][*]
Vinyl is equivalent to which digital bit-depth and sampling rate?
Reply #74
First I will discuss the MP3 and DAC related posts that many people commented on.
Yes, MP3 lossy compression actually is a serious problem and we can actually test that very simply;
Grab a CD with Red Book specifications. Now, set your iTunes to decode to AAC. Burn the disk to AAC. Now listen. Then set your iTunes to decode to WAV. Now listen.
I did this test a couple days ago because I'm looking at upscaling my entire iTunes library when I sync to a NAS server on my internal network, and I noticed a huge improvement even on small shotty speakers over my shotty DAC on my iMac.
If you can't hear the difference from this test, then you're probably a hobbyist with untrained ears. Look at the file difference. The WAV files are going to be almost 5 times the size of AAC files. The AAC encoder is dropping a whole lot more audio information than you'd care to admit. And put AAC vs. WAV aside for a second, I think we can all agree that AAC is better quality than MP3. How much more will the difference between MP3 and WAV be?

Please perform a double-blind test to back your claims up. If the difference is trivial to distinguish, an ABX of it would be a cinch to perform at <0.01 confidence.

As for the rest of your diatribe... we'll patiently wait for the responses to that as the earth spins around to the other side. You have a lot to learn and a lot to unlearn. One being that you have to concede that your professor may be as wrong as you when applying DA concepts to the real world.