[gregorio]

You're avoiding the question.

Already answered the question, so have others, so have the videos. So what’s the point in answering it again?

What if someone says oversampling to 352,800khz sounds more perfect than 44.1?

What if someone says the Earth is flat or unicorns really exist?

you're Dunning-Kruger for not getting that.

That’s funny, you don’t know what Dunning-Kruger is, do you? I shouldn’t be surprised, you don’t seem to know much about anything. I’d link to a video explaining it but you wouldn’t want to watch it! Lol

G

 

[Audiophiliac]

That's easy. You just rack up two identical tracks, one at 352,800kHz and one at 44.1kHz and see if the person can tell the difference in a blind comparison.

But if you want a prediction of how that might turn out, simply looking at the numbers will tell you. 44.1 is capable of reproducing all frequencies perfectly up to the edge of human hearing, 20kHz. There is absolutely no evidence that human beings can hear 176khz. A bat can hear up to 200kHz. That is the only animal on Earth that can hear that high. If you are a bat, you might need a sampling rate that high, but not if you are a human being.

But doesn't upsampling to 352.8 play it even more? And good point, wouldn't it then produce the sounds in the track that a bat would have heard, like mqa says they're doing?

 

[Audiophiliac]

Already answered the question, so have others, so have the videos. So what’s the point in answering it again?

What if someone says the Earth is flat or unicorns really exist?

That’s funny, you don’t know what Dunning-Kruger is, do you? I shouldn’t be surprised, you don’t seem to know much about anything. I’d link to a video explaining it but you wouldn’t want to watch it! Lol

G

Dunning-Kruger are at the bottom of that curve, to me. It's the part of life they want to be known for, compared to people who invent recording gear that works.

 

[bigshot]

I thought the argument was that they were unable to be created?

No, without a rolloff filter, frequencies above the sampling limit turn to noise, some of which might creep down in the audible range because of analog error.

44.1khz is not a huge number of samples, it's the bare minimum required to *PERFECTLY* capture 20 khz.

I corrected that for you. 44.1 is the minimum to perfectly reproduce 20kHz... It can't reproduce frequencies in the audible range for humans any better than at 44.1. You can add samples, but the sound is the same. The extra 2.05 is to allow for the rolloff filter at the top. In the early days of digital audio, analog filtering wasn't a perfect brick wall. Now we have oversampling DACs, so that extra buffer isn't really needed. But it was built into the standard, so it's still there.

Hope this helps.

 

[bigshot]

But doesn't upsampling to 352.8 play it even more?

I don't think you mean upsampling. That is just increasing size without increasing information. I think you mean "higher sampling rates".

What do you mean by "even more"? There is no "more" beyond perfect. If it reproduces a frequency range perfectly, it can't reproduce it any better. You can throw in more samples, but it still comes out the same. All the extra samples just become packing peanuts in the file size bulking it up for no reason.

And good point, wouldn't it then produce the sounds in the track that a bat would have heard, like mqa says they're doing?

Yes, higher sampling rates can reproduce higher frequencies. But humans can only hear up to 20kHz. That's been an established fact for over a century. Tests have shown that frequencies above that add nothing to the perceived sound quality of music. So you can go ahead and pack a whole bunch of extra stuff into the file, but your ears won't hear it.

In fact, not all home audio equipment is designed to deal with super audible frequencies. There's a chance that your amp will distort in the audible range if you present it with frequencies it wasn't designed to handle. There is a complete explanation of all this in the link titled CD SOUND IS ALL YOU NEED in my sig file. That is a very interesting and useful article.

Does this help you understand?

 

[gregorio]

Dunning-Kruger are at the bottom of that curve, to me.

There are no curves, they’re at the bottom of the stair steps.

It's the part of life they want to be known for, compared to people who invent recording gear that works.

Of course that’s what they want to be known for, because they make stairs. Business hasn’t been so good since the invention of digital escalators though

G

 

[Audiophiliac]

No, without a rolloff filter, frequencies above the sampling limit turn to noise, some of which might creep down in the audible range because of analog error.

Why would they turn in to noise? And what do you mean it can creep down into the audible range because of analog error? I thought it was exactly what is really was?

I corrected that for you. 44.1 is the minimum to perfectly reproduce 20kHz... It can't reproduce frequencies in the audible range for humans any better than at 44.1. You can add samples, but the sound is the same. The extra 2.05 is to allow for the rolloff filter at the top. In the early days of digital audio, analog filtering wasn't a perfect brick wall. Now we have oversampling DACs, so that extra buffer isn't really needed. But it was built into the standard, so it's still there.

So, what if someone says oversampling makes things sound even more perfect in the first place? Why are there oversampling dac's? And why does mine have 3 different NOS modes and 2, 4, yet only up to 8 OS modes to choose from? I haven't changed mine yet, I'll get all upset about the something better but not as something compared to the setting he ships them with. Because the designer knows digital is broken in the first place.

 

[bigshot]

It turns into noise because there aren’t enough samples to define the frequency fully. So it scatters bits and pieces of the ultra high frequencies around randomly.

Upsampling just pads the file out, it doesn’t add any sound quality because the frequencies covered by the sampling rate are perfectly reproduced already, and upsampling doesn’t add anything more to the sound. In order to reproduce super audible frequencies, the super audible frequencies need to be recorded at a sampling rate that will support them. Just upping the sampling rate after it’s been recorded at a lower rate won’t add anything.

Perfect means that the signal being recorded is identical to the signal coming out of the player. It can’t get better than the same without signal processing.

Oversampling raises the frequency rate so the error in the analog roll off filter doesn’t cut into the upper frequencies. Analog filters roll off, they don’t cut off cleanly, so raising it up projects everything up to 20kHz. It isn’t something you need to worry about. Just about all DACs oversample now. You don’t want one that doesn’t.

 

[Audiophiliac]

I don't think you mean upsampling. That is just increasing size without increasing information. I think you mean "higher sampling rates".

Upsampling increases the speed of the playback packets. Higher sampling rates refers to the adc chip capturing samples more frequently. The point of this thread, to me, is to complain about upsampling instead of higher sampling rates, and you're telling me anybody who thinks there should be more than 44.1 is stupid.

What do you mean by "even more"? There is no "more" beyond perfect. If it reproduces a frequency range perfectly, it can't reproduce it any better. You can throw in more samples, but it still comes out the same. All the extra samples just become packing peanuts in the file size bulking it up for no reason.

But remember the part about how connect the dots can't draw intricate custom shapes too well? And remember, we want the connect the dots to look just like each hair of the paint brush painted identically to how the original did.

Yes, higher sampling rates can reproduce higher frequencies. But humans can only hear up to 20kHz. That's been an established fact for over a century. Tests have shown that frequencies above that add nothing to the perceived sound quality of music. So you can go ahead and pack a whole bunch of extra stuff into the file, but your ears won't hear it.

If a sample comes after the actual strike of a drum, doesn't it mess with the speed, attack, decay, and detail of sounds?

In fact, not all home audio equipment is designed to deal with super audible frequencies. There's a chance that your amp will distort in the audible range if you present it with frequencies it wasn't designed to handle. There is a complete explanation of all this in the link titled CD SOUND IS ALL YOU NEED in my sig file. That is a very interesting and useful article.

So, your amp clips if it plays above 22khz. That's why you stick with cd's.

 

[Audiophiliac]

There are no curves, they’re at the bottom of the stair steps.

Of course that’s what they want to be known for, because they make stairs. Business hasn’t been so good since the invention of digital escalators though

Those people need to know they can only be perfect in 44.1. If they were, people would compress them lossily because of it.

 

[bigshot]

16/44.1 PCM and FLAC are both bit perfect and audibly transparent. With a high enough data rate, AAC and MP3 can be audibly transparent, but not bit perfect. Lossless is perfect both in the zeros and ones in the file and the sound that reaches your ears. Lossy isn't perfect in the zeros and ones, but it can sound just as good as lossless or higher bit/sampling rates.

 

[bigshot]

Upsampling increases the speed of the playback packets. Higher sampling rates refers to the adc chip capturing samples more frequently.

There is no additional data when you upsample. It just adds more of the same. Picture your music as a number like...
1-2-3-4-5-6-7-8-9-10
and then picture it as...
1-1-2-2-3-3-4-4-5-5-6-6-7-7-8-8-9-9-10-10
It's still just 1 to 10. 11, 12 and 13 aren't in there. And if you run through it twice as fast, it's still 1 for the same amount of time. That is what upsampling does. It doesn't add data, just redundant data.

But remember the part about how connect the dots can't draw intricate custom shapes too well? And remember, we want the connect the dots to look just like each hair of the paint brush painted identically to how the original did.

As I've said before, the Nyquist theory (which is the foundation digital audio is built upon) says that two samples are required to PERFECTLY REPRODUCE any single frequency. You can add more points, but it won't get any better sounding. Redundancy again.

If a sample comes after the actual strike of a drum, doesn't it mess with the speed, attack, decay, and detail of sounds?

No, because the transients that occur in music are more than 100 times slower than the time between two samples. OK, we're getting into math here, which isn't my strongest suit, but... The speed of a drum hit or decay takes up hundreds if not thousands of samples at 44.1. Remember we are talking a single sample being 1/44,000ths of a second. That is 44 times smaller than a millisecond. A fast snare drum hit from attack to peak is probably about 5 milliseconds. That is over 2,000 samples. No transient in music comes close to 1/44,000th of a second. The only thing that does is a frequency above 20kHz, which by definition would need to be have points less than 1/44,000th of a second apart? Do you follow that?

So, your amp clips if it plays above 22khz. That's why you stick with cd's.

Yes, CDs work with more amps and they sound just as good to human ears as higher bit/sampling rates do. CD Sound Is All You Need (see link below if you haven't clicked on it yet- all this stuff is explained better than I can explain it there.)

 

[VNandor]

@Audiophiliac
Here's a comparison between the same song but different sample rates. I took an audio file with a sample rate of 176400Hz. This is the actual sample rate of the recording, it's not upsampled. Keep in mind that the difference between different sample rates are the different cut-off frequency of the analog signals they can represent. So I low-passed this file with a cut-off frequency of ~20kHz. The recording still has the high resolution sample rate, I only removed the high frequencies. After that, I downsampled this file to 44.1kHz.
Here's a comparison between a slice of these file's waveform:

First of all, notice that I'm using music, not some obscure test signal. This is a very, very short slice of the music I used so you can get a better look.

There's something we have to clear up. The bottom image shows the "stairsteps" as the samples are being connected in a very simple way. This line is not what the ADC encodes. As an aside, a signal that looks like the stairstepped line would never come out of an ADC because an ADC usually puts out a PCM signal which looks nothing like that. The point of the PCM signal is not to encode information about the stairstepped signal. The output of the ADC encodes information about the continous signal which looks just like the one overlaid on top of the stairsteps.

Something you might find more interesting is the top picture. It shows what a digital audio signal is in a less misleading way. The samples are not being connected the way they are on the bottom picture. Since a digital audio signal is discrete in time, it's a better representation. Notice how the orange line that the digital audio signal encodes is smooth. Also look at how precisely the peaks are represented. They don't have to coincide with the sample points at all, the timing of the peak does not matter, it gets encoded properly. The orange line also looks exactly like the analog voltage (if you looked with an oscilloscope, obviously noone can "see" voltage) that would come out of the DAC if you fed it the digital audio signal the white dots show.

Lastly, notice how similar the top and bottom pictures are. The similarity might be unexpected to you since the music I used have a sample rate of 176400Hz and the top picture shows the 44100 signal. The reason they are this similar is because the 176400 sample is low-passed as stated before. Since it is low-passed, a digital signal with a sample rate of 44100Hz can encode it just as well as a digital signal with a sample rate of 176400Hz. The extra sample points included in the 176kHz file have no use, as the extra sample points added during the recording don't carry any extra information about the band-limited analog signal it represents.

I could also send you the tested songs so you can compare them (test if they not only look the same but also sound the same) yourself by a properly conducted blind test.

Also the last couple pages of the discussion would fit better an some other threads, including this post I guess.

 

[Lappy27]

In the 3000's, people will be getting digitized because they can't tell we don't all just look like dots right now.
And you know what's going to happen to them.

I really like MQA with my streaming devices for my hears. Voodoo or not

 

[Audiophiliac]

16/44.1 PCM and FLAC are both bit perfect and audibly transparent. With a high enough data rate, AAC and MP3 can be audibly transparent, but not bit perfect. Lossless is perfect both in the zeros and ones in the file and the sound that reaches your ears. Lossy isn't perfect in the zeros and ones, but it can sound just as good as lossless or higher bit/sampling rates.

True enough, if I made a dot on a piece of paper, people wouldn't be able to say it didn't look just like you.