[WindowsX]

What should I believe in now?

 

[castleofargh]

I don't know the context so I wouldn't take side in anything. but there is something seriously wrong when the noise reaches amplitudes bigger than the digital signal itself like in this example.

 

[WindowsX]

You can look here.

https://www.head-fi.org/f/threads/s...rd-for-hq-audio-players.755645/#post-11339716

 

[Niouke]

I find funny that the "critical listeners" find effects comparable to lower background noise, which is what they wrongly expect.

 

[castleofargh]

ok so it was about the sony µSD and their "sony style" graph without scale or signal amplitude for comparison to keep the paranoids on their toes.
noise is a thing of course. it can even at low levels impact the timing of the 0/1 switches, which is not so relevant with an async streaming system or almost anything using a buffer as the delays should not be carried on to the next reading. but if it was to reach an amplitude similar or even higher than the signal itself like in the example, I would assume that most digital streaming methods would detect an error and just stop working even with error correction protocols if the noise stays that high. I don't know if there are many digital streaming protocols that would just keep collecting the modified data as if everything was cool.

 

[Speedskater]

What's true is that the charts represent a defective system.
I can't imagine any product being that bad and there are some very poorly designed boutique products.

 

[WindowsX]

What you saw in this post was cropped from this book.

It has nothing to do with Sony Premium Sound.

 

[bfreedma]

Another wild goose chase without proper context.

Did you get the book author's permission to reproduce their work?

 

[yage]

What's in the book is a theoretical example meant to, as you note, teach students digital wireless communications theory. In practice, there's a example on that other site by atomicbob, post #571 under 'Modi Multibit: Multibit for the Masses". To summarize, he sets up a loopback using USB and SPDIF interfaces and found zero errors among 67,737,552 bits sent over the wires.

Now, when my Linux machine is under heavy load (such as compiling the kernel using 4 threads), then I will notice a drop out or two if I'm sending bits through the USB interface to my DAC. However, I think that's more an issue of latency than noise / errors.

 

[gregorio]

What should I believe in now?

The problem, as is often the case with audiophiles, is not so much about what to believe as it is about understanding what you are reading!

The page appears to demonstrate what happens if noise is of equal energy to the signal and if that's the case, yes, there are going to be severe problems. So what should you believe? You can believe that any DAC or source with a 0dB signal to noise ratio is going to have severe problems. Do you have a DAC or source with a 0dB signal to noise ratio? Have you ever seen or even heard of a DAC or source with a 0dB signal to noise ratio?

G

 

[ev13wt]

This is very simple. Bear with me here - an note I am using "dB" in a very relaxed fashion.

Noise is always present, as is the signal. How loud is the signal, how loud is the noise? For digital, you need a signal to noise difference of exactly what is needed so the dac or whatever works. More is simply overkill. It will work, or it will not work. If it works the signal is perfect. (Side note: Clocks are so precise these days, that external clocks would probably introduce timing issues - unless you are in a studio and need to sync - then be happy if it works)

Analog signal to noise is a bit different, because we can actually hear the noise. Tape hiss for instance is a great example (not of snr, but hey, bear with me) to use because many have heard it. -60dB noise floor (tape hiss) and +6db recording level, then S/N ratio is 66dB.
If you have very low musical signal, lets say a single, soft plucking guitar at a distance. Its amplitude converts to 40dB. So you have a difference of 20dB. This is a problem.
If you have a loud signal, lets say an orchestra playing full on) at 0dB (reference level), tape hiss is in all accounts "unhearable".


So again, its the dosage that matters.The ratio of the sound intensity that causes permanent damage during short exposure to that of the quietest sound that the ear can hear is greater than or equal to 1 trillion. Source: National Institute on Deafness and Other Communications Disorders (National Institutes of Health, 2008)

You may feel volume as a smoothly slideing scale of 1 to 150, but it is not true: Decibel is a logarithmic scale. Do you understand what that means? 1 isn't 100 times lower than 100.

"Twice as loud is what?"

• a loudness level of about +10 dB
• Side note: double the sound pressure means 10 times increase in power. Your speakers can do 96dB with 10 watts? you need 100 more to reach 106dB.

With this new found information: If you have a noise of 50dB, and a signal of 70dB (ie 20dB SNR) - how much louder is signal? Twice as loud? 4 times as loud? Something else?

If you have a SNR of 80dB: How much louder than the noise is the signal?

If you sit down and calculate it, you will understand.

A well complied source: http://www.animations.physics.unsw.edu.au/jw/dB.htm


Edit: I didn't see who the OP is. I made the for you relevant part of my post BOLD

 

[ev13wt]

So here we have it, and its not off topic.

Q: Are there sound quality differences in "digital" cables? If the SNR is too low, can you hear it?
A: No. If there where an issue, I could not read this forum text, because the 10000 miles of crappy tiny cable would distort the signal and all I would be reading is " sadfa a a w d asd g h th th4 gd tb!"

The question is really only: When will it stop working altogether? 20dBr? 60dBr?