[gregorio]

"We"? I certainly can't.

But you did “It is totally normal how the "old" tech loses its value, when the "new" tech comes to market”!

Irrelevant for many perhaps, but not for everyone.

It’s irrelevant how much extra we’re paying for the snake oil features, just that we do have to pay for them.

It is totally normal how the "old" tech loses its value, when the "new" tech comes to market, snake oil or not. How could DAC chips be free of this market principle?

Simple, either not have new tech, sell the same tech at a lower price, or have new tech that isn’t snake oil.

True, but Maybe 384kHz and 768kHz chips are manufactured in very low quantities meaning that almost all chips were 192kHz allowing mass production benefits.

Exactly but the same was true when 192k was the new tech and 96k was the standard and prior to that when 96k was the new tech and 48k was the standard.

I doubt that since 192 kHz supporting DACs can be sold under $20.

Again, you have that backwards. “192k supporting DACs can be sold under $20” because they do not have to pay the $5-7 premium for the newer tech chips.

Know? We are speculation and guessing here! Basically telling each other what we think about the DAC chip market today.

I’m not telling you what I think about the DAC chip market today and I’m not speculating, I’m stating the historical facts of the DAC chip market. It’s a historical fact that at one time DAC chips only supported up to 48kHz, that more expensive 96kHz DAC chips were introduced, the 48k chips dropped in price and eventually ceased to be manufactured. It’s historical fact that this pattern repeated with the introduction of 192k chips and is continuing with the introduction of 384k chips.

G

 

[71 dB]

But man, have I now typed words about a thing that I will never fully understand.

At least you know there are things you don't understand. We all start from zero understanding and work our way upwards our own pace. Often people think they understand and know more than they actually do (Dunning-Kruger effect), but that doesn't seem to be your problem. Just don't say to yourself you will never fully understand these things, because that is limiting yourself. Just say yourself you have a lot to learn. Just five years ago I understood almost nothing about music theory. It all felt "naming things" to me: Minor triad, interval of fifth etc. Then about 5 years ago I watched a Youtube video by Hack Music Theory about the new The Prodigy track and the way the music was analysed in that video caused a lot of "ahaa"-moments in my head. I suddenly understood music theory is context-related. I had been thinking music theory in a wrong way, in physical way without context were things just are. When I figured out music is about having meaning by creating context (for example, what note C means is up to the scale you are using), a massive amount of doors opened to me. Suddenly I was able to learn how J. S. Bach composed fugues and what not! So, sometimes a small amount of critical information explained to you in the way that suites your thinking can lead to massive increase in understanding.

 

[71 dB]

Again, you have that backwards. “192k supporting DACs can be sold under $20” because they do not have to pay the $5-7 premium for the newer tech chips.

G

There will always be cheaper "old tech" that has lost its value and "new tech" that is pricier because of premiums etc. It is up to the buyer to figure out which one to buy. Some people want to own the "new tech" for whatever status reasons and are willing to pay for it, while some other people may be contempt with the "old tech" and pay less for it.

You don't know this?

 

[Ryokan]

Because the other half covers the lower frequencies.

 

[gregorio]

It takes two points to create a waveform. I’m sure others will answer with a whole lot more words if you prefer that.

Not just more words but more correct words and that answers his questions. For example, it takes more than two points.

Can someone explain the Nyquist Theorem, why the highest frequency in digital audio is half the sample rate.

Over simplistically: Digital audio works on the principle of converting a continuous analogue audio signal and converting it into a number of discrete points/ordinates called “samples”. The areas between these samples can be reconstructed with a mathematical function that only has one solution for a given set of samples, which results in exactly the same waveform as our original. However, this condition is not met if we have two or less samples per wave period, in such a case our mathematical function no longer has only one solution but at least two. One solution will be our original analogue waveform but we’ll also have other solutions/waveforms (called “aliases”) that weren’t in the original and our reconstructed waveform will be a combination of these waveforms…

Also why is an anti aliasing filter necessary if we can't hear the high frequencies it rolls off?

To avoid the above scenario of our original waveform distorted by aliases, all we have to do is remove any wave periods that span 2 or fewer sample periods. In the case of a 48kHz sample frequency, any audio frequency less than 24kHz must span more than two sample points. So we use a filter that removes all 24kHz and higher content (even though we can’t hear it), so there are no aliases (which are within the audible spectrum) to distort our original waveform.

Please note that there are various different other simplified ways of explaining the above. So you might get quite a different explanation that is also correct.

G

 

[Ryokan]

Over simplistically: Digital audio works on the principle of converting a continuous analogue audio signal and converting it into a number of discrete points/ordinates called “samples”. The areas between these samples can be reconstructed with a mathematical function that only has one solution for a given set of samples, which results in exactly the same waveform as our original.

44.1khz samples at 44100 times a second yet there are areas between these samples? or the area between sampling start and end points?

 

[gregorio]

There will always be cheaper "old tech" that has lost its value and "new tech" that is pricier because of premiums etc. It is up to the buyer to figure out which one to buy.

Obviously it’s not! The tech we need is sample rates up to 48 (as anything beyond that is snake oil), tech that allows us to connect that DAC to our current sources, say USB2 and higher, Ethernet, SPDif or others and tech that implements some other features/functionality we may find useful. So “figuring out what to buy” doesn’t make any difference if you can’t buy it and instead you have to buy one that includes snake oil!

You don't know this?

I do, that’s the whole point! Even with relatively “old tech” you can’t avoid buying snake oil, only with discontinued tech can you avoid the snake oil but it doesn’t have the modern functionality and it’s discontinued so you can’t buy it anyway!

Is this really so difficult a concept to grasp?

G

 

[gregorio]

44.1khz samples at 44100 times a second yet there are areas between these samples?

Of course, there must be. A continuous signal has an infinite number of points between the sampling points, although of infinitely small duration.

You mean the area between sampling start and end points I guess.

There isn’t a sampling start and end point, there’s just a single sampling point (or rather a series of single sampling points).

G

 

[gregorio]

This graphic from Wikipedia might help:
The dots on the two left images are the sample points. Notice they’re individual points with no start or end. The values of those sample points are the same in both the top and bottom left images all the time. The frequency of the sine wave is shown in the top right image.

Notice that the two left images are identical until the sine wave frequency reaches the Nyquist Point (half the sample frequency, fs/2) at which point the bottom left image diverges from the top left. This divergence is caused by now having 2 or fewer sample points per wave period and shows the alternate solution to the same sample points. We now have an alias (in the spectrum below the Nyquist Point) and what would be reconstructed is a combination of both the top and bottom left images, the two frequencies displayed in the bottom right image.

An anti-alias filter set to remove all audio content at and above the Nyquist Point will result in the top and bottom images always being identical. Ie. Our sampled input being identical to our output.

Not sure if this helps?

G

 

[vergesslich22]

At least you know there are things you don't understand. We all start from zero understanding and work our way upwards our own pace. Often people think they understand and know more than they actually do (Dunning-Kruger effect), but that doesn't seem to be your problem. Just don't say to yourself you will never fully understand these things, because that is limiting yourself. Just say yourself you have a lot to learn. Just five years ago I understood almost nothing about music theory. It all felt "naming things" to me: Minor triad, interval of fifth etc. Then about 5 years ago I watched a Youtube video by Hack Music Theory about the new The Prodigy track and the way the music was analysed in that video caused a lot of "ahaa"-moments in my head. I suddenly understood music theory is context-related. I had been thinking music theory in a wrong way, in physical way without context were things just are. When I figured out music is about having meaning by creating context (for example, what note C means is up to the scale you are using), a massive amount of doors opened to me. Suddenly I was able to learn how J. S. Bach composed fugues and what not! So, sometimes a small amount of critical information explained to you in the way that suites your thinking can lead to massive increase in understanding.

I want to try to say this: Thank you, it's true that I need to stay on track learning such things. And I think it often comes down to accepting that something dominates the real world and what's happening there. E.g. mastering is done with speakers, the outcomes of systems like filters are estimated with maths and often with closed formulas, and mathematicians use their notation and also typical diagrams and explanatory pictures for good reasons. I'm not fast at maths, so, it is what it is. Quite often I cannot get over the fact that there was a time period when math statements were written out as long sentences, which looks easier for me, but not for the majority of people who use it professionally. This all plays into understanding filters and certain methods.

With composition I'm too about to learn about it. My previous view has been that I just couldn't find what would dominate the feelings of listeners when it comes to harmonies - and rythm, and tuning of instruments. I just started to make music and was convinced that there is no way to find out that some thing would lead the top causes of feelings that music triggers in listeners. Also the way that it blends into cultural preferrences and early childhood learning was sometimes discouraging for me, I didn't know what I actually wanted to achieve.

My view today has changed in some respect. I see the impressive acomplishment that the composers of the likes of Bach have achieved. TBH I was quite often hindered by the fear that I might tarnish the fun of music making with any process that is systematic and therefore would also show me how much work is to be done, e.g. by entering all notes and pressing all buttons, all the decisions that are more the result of math-like thinking and less the result of just fun and some experiments. It's like some question .. like do you want this to be a hobby, an extensive hobby or even some sort of career. And I have not decided yet how big the hobby shall be, but I don't have to, currently. I have found some personal art form and will see what it's good for at some point in the future.

 

[castleofargh]

@Ryokan , I like that video on aliasing, maybe you will too.
The one after that is on quantization and bits. Both should help explain many things about digital sampling and reconstruction of the analog signal.

 

[Corti]

So, sometimes a small amount of critical information explained to you in the way that suites your thinking can lead to massive increase in understanding.

Well said.

 

[Ryokan]

So if we don't have a sample rate at least twice the highest frequencies a computer won't know how to connect the sample points correctly? Twice frequency gives more waypoints?

 

[texx2818]

It's a videogame, child.

 

[Corti]

It's a videogame, child.

Nope, it's a meme, showing you being completely laughable.