Reasoning towards a desired result, we all can do it.
Here is where i part this thread.
Alchemists making digital encoded music better in the digital domain can take over 
.
The sun turns around the earth. I see it coming from the east every morning and going down in the west every evening.
No. It is proven that the earth turns around the sun and not otherwise. That is common knowledge.
Might be, but I can see it clearly every day. It is the sun that moves.
And the discussion continues endlessly…
3 Likes
Data bits are independent of the storage media or delivery method. They are information and can be stored as an electrical charge, a magnetic pole orientation, or a line on a piece of paper. With error correction the the data integrity will be the same regardless of the medium with which they are stored or how they are delivered. How the data are processed and converted is another story.
4 Likes
So if not visible, it cannot be known? How then we know electrons exist, or for that matter, DNA or any other microscopic matter? If electron flow is not analogous to water flow, then where does jitter comes from, as well as all kinds of digital noise?
it was with reference to the analogy of water… but yes anything we can’t detect with our human senses begs the question of existence.
1 Like
It’s amazing how little people know about digital transmission. And only one person mentions Jitter so far?
First, we can all agree on what bits are; On or Off. Some of you are burying your head in the sand at that point. Lets look just a little bit deeper:
Yes, I’m over simplifying, but this is just to get the difference across
When a computer sends data (bits) to a printer (for example), it sends it in packets. Lets just look at it like this: X0X0XX00 0X0X00XX
Now if the receiving end gets X0X0XX00 0X0X00** the receiving end will go “crap, something didn’t come across as expected, please resend”. and the computer will keep resending till the receiving end goes GOT IT! Meanwhile, all the other data is still flowing - and the receiving end simply puts the missing packet in the right place and your print job comes out.
Now, music bits are completely different, BECAUSE THEY HAVE TIMING.
Now, instead of packets of data that can be received out of order and reconstructed, we have a steam of bit (yes, still On and Off), but it is a continuous steam with timing.
So using my XO “bits” lets say it looks like this:
X 0 X 0 X X 0 0 0 X 0 X 0 0 X X
Now on the receiving end it gets:
X 0 X 0 X X 0 0 0 X 0 X 0 0 X X
Same thing bits are bits and nothing else matters!! Right?
No…look closely. The “timing” is off. A few of those bits arrived just a tiny bit out of time. BUT BITS ARE BITS!! Yes, all the bits are there…so you are right, but you’re ignoring the timing. That’s called Jitter, which is distortion on the digital level.
Yes, yes, you can argue semantics and details and talk about buffers and caching and timing correction…but, that just proves my point…if “bits are bits” and that ends the discussion, then none of that correction would matter.
So yes, bits are bits, and everything else matters.
1 Like
So, funny and ironic thing…in my example above, I used 3 spaces instead of 2 spaces in the music streaming to show the “timing” being off…and the forum corrected it and made my incorrectly timed transmission exactly match the orignal…kinda funny, but I hope my point still gets across.
1 Like
“Music bits”, as you put it, are no different to any other bits when they are carried in Ethernet or USB packets. They only have timing insofar as the sender and receiver agree on the sample size and rate; the difference in arrival time between the bits is in general completely unrelated to the audio sample rate.
4 Likes
Your story is true for SPDIF signals, not for ethernet transfer, they don’t work the same. Yes there is also jitter in ethernet connections but jitter here is refereed to packets not arriving in time to be able to fill the buffer and that will cause delays and in case of audio streaming dropouts. Now Roon buffers up to the whole song when streaming it from Tidal or Qobuzz. The data isn’t streamed in real time, it is played from your harddrive actually because that where the data is stored. From Core to Endpoint Roon also buffers but the buffer is much shorter, something like 3-5 seconds of audio. This is enough because most of the time local nwetworks are more stable then internet connections. At the moment the data leafs the buffer in the endpoint that is where timing gets involved, and that where the jiiter you are referring to comes into play, not in the stages before that. I am not in the bits are bits camp but you do have to realize that not everything digital automatically functions in the same way. TCP/IP and SPDIF are very different kind of signals
5 Likes
LOL hilarious
1 Like
True, when they are being processed, that’s when errors are introduced. Stray currents, currents generated by auxiliary circuits etc, the so-called digital noise.
It seems many people think “bits are bits” means that nothing besides the bits themselves matter in a digital playback system. That is of course not the case. But saying that “everything matters” is also wrong. In a functioning digital playback system, there should be zero data loss when bits travel between the source (some kind of storage, local or remote) and the DAC, just as there is zero data loss when they travel from one computer to another through a network (local or remote). Any wired or wireless connection that functions properly can achieve that and should be fast enough to accommodate the highest audio bit rate. Then, if same bits are presented to a DAC, regardless of how the bits got there and through which input they enter the DAC, and if the DAC’s specifications exceed human hearing capabilities, then the output will sound the same every time. In other words, you just need a high quality DAC. Those are a commodity these days. That is hardly “everything”. Expensive transports, servers, clocks, cables, de-jitters etc. only work on a psychological level.
3 Likes
This article may or may not add anything to this discussion:
Just what we need, more marketing talk.
4 Likes
If bits are lost you can hear clicks, so that is obvious.
So yes all bits arrive at the DAC if you don’t hear the clickerdy click sound in the music.
When those audio bits (in the form of an electric signal) arrive at the DAC they are not alone, they are accompanied by electrical noise. It is the amount and the form of this noise that matters A LOT.
(TOSlink is an exeption on this rule, but hat input has other problems)
And then there is the DAC itself.
In this part a lot goes wrong in your assumptions IMHO.
How they got there matters because it introduces electrical noise. This can be from mild to very aggressive and therefor has an influence. How the DAC eliminates / handles this electrical noise is totally the responsibility of the designer. On top of that is buffering (under run / overrun) internal noise of power supply and again design choices. And the list goes on and on.
Specifications say diddly squat about the sound of a DAC. Worse measurements can sound better and vise versa. Measurements can help you a little but not very much.
No they are not. Again my opinion.
A good DAC is a true rarity, not a commodity.
But maybe my standard about what a good sounding DAC differs from others opinions on that matter. I accept that.
Do we have to go back to basics? Anyone with some knowledge about digital interfaces know that there is a generous range of voltages assigned to the two logical levels. For example, in TTL, anything below 0.8V will read as a zero bit, and anything above 2V as a one bit. In CMOS, below 1.5V is zero and above 3.5V is one. You’d have to have a huge amount of noise to go outside these margins, and if you do, there are correction algorithms that can recover occasional bit flips. That’s why digital is superior to analog here. The DAC is perfectly capable of internally reconstructing the original bits coming from storage, before doing its internal magic of analog conversion. As for the conversion itself, all the issues you mentioned related to supply noise, buffering, jitter reduction etc. are simply showing in the measurements. If it measures well when used as designed (plugged into a regular outlet and using its internal clock), then it’s good. If not, it simply colors the sound. You might like the colors, but you’re not listening to the original material anymore.
As for commodity DACs, there are units (e.g. Topping, SMSL) in the few hundred dollar range that are pushing the limits of sound analyzers. At the same time, there are boutique DACs in the thousand dollar range that reinvent the wheel so that they can use buzzword marketing terms, butcher the sound and advertise it as “signature”.
3 Likes
Yes, they are a commodity, just like good amplifiers. We could go back and forth on this forever, in the manner of some other threads here. But I think it’s more interesting to try to tease out the reason these misconceptions keep surfacing.
In the 1980s, when digital music was still in its infancy, lots of the folks who were building conventional analog audio equipment tried to build both CD players and integrated amplifiers capable of rendering digital inputs from high-end CD players. Many of them were very badly designed, both because the designers were not well-versed in digital techniques and problems, and because the packaged D/A converters available to them were relatively crude. Back then, it was indeed easy to buy very expensive but very bad equipment. And when DACs started to be sold as separate components, many of them suffered from the same problems. A good DAC was a rarity, in 1985.
But the thing about engineers is, they like to solve problems. They like to solve new problems, they like to solve old problems. And sometimes they like to solve problems even if their sales-oriented bosses tell them it wouldn’t be cost-effective. And during the 36 years since 1985, they’ve found solutions to most of the problems with simple audio D/A conversion. So the problems you refer to, the problems with electrical noise and power supplies and jitter, those are problems the analog past had, fumbling its way into the digital world, and in the present can be removed from consideration by simple state-of-the-practice engineering.
Note that I say state-of-the-practice, not state-of-the-art. That means that the processes and techniques for doing this are well known, and taught in engineering school. You don’t need some Zen master of design to get there. You need someone who knows what they’re doing, who is familiar with today’s engineering techniques, not yesterday’s.
Now, a lot of folks still like to use old audio equipment, or new equipment based on old designs, or new equipment designed by old designers not terribly familiar with modern techniques. A lot of people like to assemble frankensystems from a myriad of disparate components, just as we did, gleefully, 50 years ago. So, sure, the old problems can carry through in that old technology, in those old practices. No question about that. But there’s no particular reason to suffer with those problems, aside from an understandable nostalgia for a bygone age. Heck, I sometimes like to read by the light of a kerosene lantern my grandfather built. But when I listen to music, I use a transparent system, built from commodity components by young designers who simply assemble things the right way. It’s easy.
5 Likes
Are you sure about this? I thought this wasn’t the case and thus the reason we don’t see waveforms for streamed music versus files that are actually in your local library.
Nope. Great DACs, great amps, and great line stages are not commodities. If they were, they would be everywhere and they are not. I know because I have looked…