Why 24 bit audio and anything over 48k is not only worthless, but bad for music.

[theveterans]

How do you digitally make a wave?

And I'm not sure what the point of your original question was; square sound waves do not exist so it is a non-sensical question. Sound is a harmonic phenomenon (in the linear range of normal sound pressure levels), so there are only sine waves in sound. Nothing else. No square sound "waves", no triangular sound "waves" etc.

The second you pump the output of a function generator's "square" function into a sound transducer, it becomes a bunch of superimposed harmonic (sine) sound waves. That's how a compressible gas medium works.

Edit: for the avoidance of confusion, the above is directed at @theveterans, not @VNandor. I'm sure VNandor realises that, but visitors to this forum might not.

Audacity or Adobe Audition

My point is using a square wave input, you can actually see differences between DACs (some will have a more ripple aka a bunch of sine wave overlaps using sinc function). But since you can't hear beyond 20 KHz, the 7KHz square wave will sound exactly like a 7KHz sine wave.

This is where anything above 48 KHz to even (16fs) 768 KHz and you can see differences (not in bandwidth since nobody can hear beyond 20 KHz) but on an illegal input like square waves

 

[2leftears]

Audacity or Adobe Audition

My point is using a square wave input, you can actually see differences between DACs (some will have a more ripple aka a bunch of sine wave overlaps using sinc function). But since you can't hear beyond 20 KHz, the 7KHz square wave will sound exactly like a 7KHz sine wave.

This is where anything above 48 KHz to even (16fs) 768 KHz and you can see differences (not in bandwidth since nobody can hear beyond 20 KHz) but on an illegal input like square waves

I think you missed my point; in the digital audio domain waves do not exist, only discrete samples of zero duration defined on discrete time points. But never mind that, consider it pedantry on my part if you wish

So what you really mean is to mathematically (or programmatically) create a series of piece-wise constant digital audio samples that could never have been recorded in compliance with the finite bandwidth digital sampling theorems as applied to the (human audible) audio bandwidth?

Why would I care if different audio DACs handle that situation differently? We are talking about audio DACs (i.e. for music) designed specifically to reproduce (audio) bandwidth-limited content from appropriately bandwidth-limited sampled digital content. We are not talking about DACs for use in 100MHz bandwidth function generators.

Maybe I also missed your point?

 

[theveterans]

I think you missed my point; in the digital audio domain waves do not exist, only discrete samples of zero duration defined on discrete time points. But never mind that, consider it pedantry on my part if you wish

I can have a quantized generated square wave in Audition or Audacity and let the DAC handle that in their own designed way. Depending on how the DAC handles it, you can have different results which can affect different tonal balance (depending on how much precise the analog square wave is reproduced). In case I'm making this stuff up, there's literally so many DAC square wave measurements out there in the internet BTW. Also, this is where the original topic of

Why 24 bit audio and anything over 48k is not only worthless, but bad for music​

My point is feeding the DAC that's 768 KHz or even DSD512 ISN'T worthless but has a measured effect on tones that are literally within audible range

 

[Davesrose]

My point is feeding the DAC that's 768 KHz or even DSD512 ISN'T worthless but has a measured effect on tones that are literally within audible range

But if harmonics that are above 20 kHz gets included in the recording (the ADC), why does a DAC need to be able to produce over 44.1 kHz?

 

[VNandor]

Audacity or Adobe Audition

My point is using a square wave input, you can actually see differences between DACs (some will have a more ripple aka a bunch of sine wave overlaps using sinc function). But since you can't hear beyond 20 KHz, the 7KHz square wave will sound exactly like a 7KHz sine wave.

Have you ever actually tried this and listened to the result or is this one of your assumptions you never bothered to check again? I generated a 7kHz square wave with your suggested settings and it sounds nothing like a 7kHz sine wave. I can share the result if you want, the difference is not subtle, no golden ears or hifi equipment is needed.

 

[Davesrose]

Have you ever actually tried this and listened to the result or is this one of your assumptions you never bothered to check again? I generated a 7kHz square wave with your suggested settings and it sounds nothing like a 7kHz sine wave. I can share the result if you want, the difference is not subtle, no golden ears or hifi equipment is needed.

I think I found a video that already goes over this:

 

[theveterans]

Have you ever actually tried this and listened to the result or is this one of your assumptions you never bothered to check again? I generated a 7kHz square wave with your suggested settings and it sounds nothing like a 7kHz sine wave. I can share the result if you want, the difference is not subtle, no golden ears or hifi equipment is needed.

Yes. They sound the same. Measurements of 7KHz FFT of a square wave support what I hear in this case. Your DAC is broken in this case

 

[danadam]

I generated a 7kHz square wave with your suggested settings and it sounds nothing like a 7kHz sine wave.

Frankly I'm still not sure what settings he has in mind. It doesn't seem like he clearly answered if he means a square that is properly band-limited or not.

 

[bigshot]

I think he said he used a signal generator. Where does the DAC come in? This is out of my ballywick perhaps.

 

[Davesrose]

I think he said he used a signal generator. Where does the DAC come in? This is out of my ballywick perhaps.

If you're referring to my video, it's in the second segment (setup). He mentions he created the sound digitally (so there's a DAC). Anyway, I think there's been crosstalk with different terms. I think now understand you and 71dB's points as far as sampling with digital audio, (as far as waves never really being a perfect square in digital audio: why there's anti-aliasing). Your statement that "square waves are pretty much irrelevant in music" is probably the best summation. And all this is moot if it seems the underlying premise is we need sampling above 44.1 kHz.

 

[Ryokan]

I'm trying to understand the Fourier Series, but many videos are aimed at university level folk who already have a good understanding of the math terminology.
This one has been more helpful to me,

 

[bigshot]

No I was talking about VNandor’s comment.

 

[71 dB]

I think there can be some miscommunication and possibly me not understanding. I thought no square signal meant the representation for what a sound wave is: a sine wave that has height (amplitude) and width (frequency). The analogy I was drawing was with vector art, we can have a representation of a curve with just a minimal set of data points. In which you can reproduce that curve with same curvature at any size. With digital audio, I understand reproduction of a sound wave includes sampling discrete measurements (steps or "square"). That it's not "illegal" to have the waveform setting to be "square" or "alias/no alias" in software.

The last few pages have been surreal: People suddenly struggle to understand each other.

Anyway, perfect square waves are "illegal" in digital audio, because they don't obey the band-limitation requirements. Illegality here doesn't mean prison sentences. It means the analog signal after DAC is not what was meant. Splines are something completely separate.

 

[VNandor]

Yes. They sound the same. Measurements of 7KHz FFT of a square wave support what I hear in this case. Your DAC is broken in this case

My DAC is fine. Here are the results of your suggested settings, the square and the sine. The square wave will sound nothing like the sine in any of the DACs, including yours.

Can you share the 7kHz square wave you generated with adobe audition?

Frankly I'm still not sure what settings he has in mind. It doesn't seem like he clearly answered if he means a square that is properly band-limited or not.

He made it clear he does not want any ADC involved, specifically said full scale square wave with Audacity or Adobe audition. This suggests he wants no low-passing involved.

 

[2leftears]

My DAC is fine. Here are the results of your suggested settings, the square and the sine. The square wave will sound nothing like the sine in any of the DACs, including yours.

Uhm, both of those sound like a 7kHz fundamental with a lot of harmonics? Neither sound like a 7kHz sine wave. Are you sure you didn't mix up some files here?

He made it clear he does not want any ADC involved, specifically said full scale square wave with Audacity or Adobe audition. This suggests he wants no low-passing involved.

Yup, that's what I figured, no low-pass filtering. That means supplying an audio DAC with improperly constructed data it was never designed to handle.