• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Is Digital Audio Transmission Analog? [video]

Soundguy: As everyone surely agrees an FFT Is digital.
FFT - may be digital.
...
Agreed... but worth expanding.
The FT (Fourier Transform) is not discrete or digital. It transforms the time domain to the frequency domain. Each domain is inherently continuous/analog.
The DFT (Discrete Fourier Transform) is a finite version of the FT for sampled functions, so yes, it is discrete, not continuous.
The FFT is an optimized algorithm for computing the DFT. So yes, the FFT is discrete, even though the Fourier Transform in general is not.
However, even the FFT is in some sense continuous, because it's a finite sequence of functions (which makes it seem discrete), yet each of those functions (cos or sin) is continuous.
But when it's applied in digital computers, it is of course discrete.
 
FFT - may be digital.

But also FFT - if it is anything at all in this discussion - is just an analogy for how the brain is working, and I - for one - am somewhat doubtful it is an accurate one. For one thing - no one on the planet actually understands in detail just how the brain is working.

One thing though we can be certain of - the processing in the brain in no way creates numbers (digits) of any sort and hence is not digital. It has evolved out of the primordial swamp and to pretend it is working in any way like a designed digital computer is laughable.
it is laughable to imagine a computer using fingers to count or actual numbers (numerals) to operate. Digital in the context of computers doesn’t mean numbers like you would write using typical Arabic symbols or digits of your hand to count. Generally “digital” in this context refers to the most common computer design having an underlying architecture of two states - on and off or 1 and 0 or True and false or however you wish to label the two states - these are example of digital systems which is very different from analog. Neurons have two states also - either on or off. DNA nucleotides are a code with 4 options per nucleotide (or bit). The fundamental underlying human architecture is “digital” similar to that of digital computers even though chemistry is used rather than electric circuits.
 
... Digital in the context of computers doesn’t mean numbers like you would write using typical Arabic symbols or digits of your hand to count....
Really??? OMG in all my years as an electronic engineer, I never knew that :rolleyes:
 
View attachment 223349
Never bites the hand, that feeds him.
To Dark - Ever hear of digital error correction, Hamming codes, etc? All that stuff you claim to be an issue disappears (it is fixed in the DA converter and there is much error correction data and redundancy included in the digital encoding to allow for this) . The fact that digital data has to travel as analog is why all that error-correction is necessary in the first place! Read up on digital error correction, and then retract all your nonsense, please.
 
Last edited:
... Digital in the context of computers doesn’t mean numbers like you would write using typical Arabic symbols or digits of your hand to count. ...
Really??? OMG in all my years as an electronic engineer, I never knew that :rolleyes:
What if you scratch the numbers into clay tablets, like this?
What's particularly interesting about tablet YBC 7289, is they computed the square root of 2 more accurately than is possible through physical measurement, which means they had a mathematical method for deriving it.
 
Is a single bit error in a 16-bit word 44.1 KHz sample going to cause glitches? Are there more significant bits than others in D/A conversion to be noticeable?

There could be errors that do or don't get noticeable. It doesn't have to imply long cable runs to get a degraded signal to cause errors.

Or does a -130db noise line in jitter test mean that it's bit perfect?
 
Is a single bit error in a 16-bit word 44.1 KHz sample going to cause glitches? Are there more significant bits than others in D/A conversion to be noticeable?

There could be errors that do or don't get noticeable. It doesn't have to imply long cable runs to get a degraded signal to cause errors.

Or does a -130db noise line in jitter test mean that it's bit perfect?
Bit error rates in ethernet and USB cables are extremely low. So low in a domestic environment as to be discounted completely.
 
Is a single bit error in a 16-bit word 44.1 KHz sample going to cause glitches? Are there more significant bits than others in D/A conversion to be noticeable?
Maybe, and yes. Each sample is a 16-bit value representing the amplitude of the sound wave at that point in time. If a single error hits the LSB (least significant bit) it will not glitch; it will be indistinguishable from dither. If a single error hits the MSB (most significant bit), it will change the total amplitude by half of full scale (higher or lower). If the true value was a 0 and the error flips it to 1, then you get something similar to the 1-bit spike often used to show differences in linear vs. minimum phase filters.

Note on LSB bit errors: the HDCD protocol was based on manipulating the LSB of samples to encode extra information. With DACs that don't support HDCD, this is so far down in the noise that it is transparent.

Or does a -130db noise line in jitter test mean that it's bit perfect?
Under normal circumstances, playing a CD or other digital media is bit perfect. The above error scenario is hypothetical. Jitter errors do not imply that the data wasn't bit perfect. A DAC can be bit perfect and still have jitter.
 
Thanks. What I mean is that there could be a max amount of isolated errors per second due to jitter or line noise that could get unnoticeable until a certain threshold.

Even on a MSB, a single error in a sample would go unoticeable since it is so short, 0.02 ms.
 
Soundguy: As everyone surely agrees an FFT Is digital.

No. FFT performs multiplication between the input signal and weighting function in the time domain. Digital FFT analysers work on the principle of the so called Discrete Fourier Transform, DFT.


 
Last edited:
I see a lot of comments about Roon servers/cores "needing" to be separate from the audio system, and feeding digital audio to a DAC via some sort of endpoint/streamer that is located with the audio equipment. The comments are related (I think) to computer related noise. Based on what I heard in the video, a USB signal direct from a Roon Server (or other music serving device) should be fine, though it might make sense to separate the server from the audio system by a few feet (maybe 5+). Am I missing something?
 
No. FFT performs multiplication between the input signal and weighting function in the time domain. Digital FFT analysers work on the principle of the so called Discrete Fourier Transform, DFT.
...
I can see the DFT / FFT either way, discrete or continous.

They apply to sampled waveforms and even have "discrete" in their name. And the result of applying the DFT or FFT is a finite set of wave functions (sometimes represented as complex numbers). A finite set must be discrete. However, each of the wave functions in the set (sin or cos) is continuous. A sum of continuous functions, is a continuous function.

In this sense, the FFT returns the set of waves that when summed, produces the continuous waveform whose sample points were given as inputs.

However, one could still say the FFT is still discrete because it's not actually returning the wave functions. Instead, it returns the frequency, amplitude and phase of each wave. Those are just 3 numbers, not the continuous function they represent. So that is discrete.

Of course, everything becomes discrete when you compute and store it in a digital computer.
 
I see a lot of comments about Roon servers/cores "needing" to be separate from the audio system, and feeding digital audio to a DAC via some sort of endpoint/streamer that is located with the audio equipment. The comments are related (I think) to computer related noise. Based on what I heard in the video, a USB signal direct from a Roon Server (or other music serving device) should be fine, though it might make sense to separate the server from the audio system by a few feet (maybe 5+). Am I missing something?

There is a general trend of separation of functions in audio equipment on the premise that this results in a cleaner audio signal. So, where one integrated box would do just fine, two are being introduced; if two would be sufficient, three or four are needed; and so on (examples: a digital "streamer" between the digital server and DAC, phono-preamp, line preamp, power amp, monoblock amps, external power supplies, power conditioners, etc.) The goal, of course, is to sell more boxes and make the system look more "sophisticated" and "exotic" so as to support owner bragging rights about "performance superiority" and "uniqueness." But as @amirm 's and other reviewers' measurements have clearly demonstrated, for the most part there is no need for this box proliferation because there exist inexpensive integrated models that perform superbly.
You would want to separate boxes by some distance only if you believe there is electromagnetic interference (not carried by the interconnecting cables) between them, for example, a magnetic field from the mains transformer in one box causing audible hum in the other one. Well designed equipment is immune to this type of interference as well as the one carried through interconnects, even "digital" ones.
 
There is a general trend of separation of functions in audio equipment on the premise that this results in a cleaner audio signal. So, where one integrated box would do just fine, two are being introduced; if two would be sufficient, three or four are needed; and so on (examples: a digital "streamer" between the digital server and DAC, phono-preamp, line preamp, power amp, monoblock amps, external power supplies, power conditioners, etc.) The goal, of course, is to sell more boxes and make the system look more "sophisticated" and "exotic" so as to support owner bragging rights about "performance superiority" and "uniqueness." But as @amirm 's and other reviewers' measurements have clearly demonstrated, for the most part there is no need for this box proliferation because there exist inexpensive integrated models that perform superbly.
You would want to separate boxes by some distance only if you believe there is electromagnetic interference (not carried by the interconnecting cables) between them, for example, a magnetic field from the mains transformer in one box causing audible hum in the other one. Well designed equipment is immune to this type of interference as well as the one carried through interconnects, even "digital" ones.
Thank you. This is what I have always believed to be true, but wanted confirmation on this particular issue. Plan is for Roon Server to run on a Mac Mini and run a USB audio stream from it to my DAC (MiniDSP Flex).
 
I see a lot of comments about Roon servers/cores "needing" to be separate from the audio system, and feeding digital audio to a DAC via some sort of endpoint/streamer that is located with the audio equipment. The comments are related (I think) to computer related noise. Based on what I heard in the video, a USB signal direct from a Roon Server (or other music serving device) should be fine, though it might make sense to separate the server from the audio system by a few feet (maybe 5+). Am I missing something?
My roon server directly connects to my DAC.
 
Back
Top Bottom