Version 1.8 improved sound quality

Yeah, this is another part of the problem. These “highly resolving systems” are always some Frankenstein construct built from various expensive pieces all higgledy-piggledy connected by other unnecessarily expensive pieces. Stitched together by some enthusiastic amateur with no knowledge of system integration, but suffering from dollops of self-regard and a healthy dose of the Dunning-Kruger effect. Little wonder some mismatched part of their “ultimate system” is picking up noise from somewhere!

Meanwhile, the people with professionally integrated systems can’t hear the same problems the poor roll-it-yourself folks do. Go figure!

The real issue, though, is that the folks who regard a pastime of connecting together various things that have no rational reason for existing, like super audio Ethernet switches, then seeing if they can “hear” a “difference”, as a “hobby”. No worse than collecting bottle caps, I suppose. To me, that seems a similar waste of time. But these folks swarm on other audio fora, and when they come to the Roon forum, they think it’s a collection of fellow sufferers.

But Roon is really about simplicity, sophistication, and integration. It’s the antithesis of discrete audiophile connect-o-fests and various hobbyist music-player software projects. So they’re supprised to find folks on it who can see that the emperor has no clothes.

9 Likes

I think my perspective comes the fact I should never even own the setup I have. I lost both of parents at the beginning of the year and used some of the inheritance to buy the mscaler/Dave as a way of dealing with the grief. I feel like any day someone is going to come in and repossess it. I feel I got to get as much time in listening before it’s taken away. It’s an irrational fear.

Sorry to get sidetracked.

3 Likes

Look, making sure you are dealing with bit perfect data is a great way to eliminate the variable of software manipulated music data. Anyone should realize that the fewer variables involved in a problem the more likely you are to identify the real problem. As I said before, I use HQPlayer and NEVER send bit perfect data to my DAC for my every day listening. I only go to the bit perfect setup when I am testing certain things.

Now, if the DAC receives bit perfect data from a device using Roon RAAT or whatever protocol SqueezeLight uses, what could account for the difference in sound quality? If the music is not skipping or pausing coming from the DAC, we know the data got there and in time. What else comes along with the music data besides the zeroes and ones? Electrical noise? Clock phase noise?

That’s all fine and good. It also explains though why the difference is greater to me. Despite the smears of the naysayers, there is a tremendous amount of repeatability and predictability in outcomes when assembling a system.

I would expect the $20 Amazon cable to sound no better than the stock one. The stock BNC cables that came with the HMS are terrible though and it doesn’t take a lot of money to beat them. I think you will be shocked by these for example:

And no, the PowerAdd battery doesn’t do as much as what they claim as far as stopping RF.

Seems like getting sidetracked. I have replaced bnc cables with $40 Blue Jeans cables. The point of the ubs cable was I was expecting any kind of difference, but I hear none. The cable was more of a test to see if worth it to buy more expensive ones. And as for as RF from the mscaler, I tried placing the mscaler further apart vs then normal 3 inches below the Dave and also hear no difference whatsoever. So while I’m open accepting differences, I test them myself if just a matter of time or very little money.

As said, I actually do hear a difference between RAAT and Squeezelite. I’m now currently undecided which I like better.

Just wanted to add the Poweradd claim was from Rob Watts himself, the guy who built the mscaler and Dave.

Yeah same guy who shipped the Blu2 with a severe RF problem such that he had to make a change in the HMS. I think you and I may have debated this over on HeadFi. I’m a huge fan of what Rob has done with the HMS and his DACs. Those of us who have experimented with other methods just hear things differently, I guess.

Anyway back to the topic of roon sound quality. After more listening I’m back to saying roon RAAT sounds better than Squeezelite. Little bit more depth and air to soundstage.

Can take any discussion of Chord sound quality over at head-fi but think it’s been played out as well.

This has to be the funniest thing I’ve read all day. Yeah, the program that takes at least three clicks to get anywhere, and then one has no idea whether they’ll be scrolling up or down, or how to find all that’s been newly released by Qobuz or Tidal without deep diving a dozen different genres and ‘view all’ tabs. Let’s not even talk about all the “my $$$ Nucleus can’t be found by my remote” posts (or how my ROCK database has to be reset every time there’s an update).

There’s also those who swear that turntables are the best - and simplest. Except for the phono amp, the cartridge and tonearm upgrades, the special shelf it needs to sit on, the record cleaning machine, and of course the cost of the vinyl itself. Yes, one can plug and play a turntable into a receiver, and one can just stream via bluetooth from their phone, but until you’ve heard MY system, please don’t comment on how your simpler one is better, or that I’m any more of an amateur than you or anyone else on this forum is. But you’ve been condescending to myself and many others before, and I’m sure it will happen again, and again, and again…

Funny. Roon’s UI has nothing to do with sound quality. Though I’m not one of the Roon fanboys, I do readily admit that Roon has an elegant simplicity about it when it comes to getting the music to where you want it to go. On a broad array of devices no less. So credit where credit is due: this they do well.

As for turntables: scoff all you want. I am still very much in love with vinyl.

Most of the minutiae you mention have a solid base in physics. Cartridges contain moving parts that are sensitive to vibrations. Any vibration added to the movement of the cantilever will change the sound. Just tap on the cartridge and you will hear the tap over your system because the vibrations will result in a different modulated current. The same goes for the arm. It too can add vibrations.

Ditto for a shelf with the added caveat that a turntable that is not level will cause the stylus to ride unevenly in the groove, thus not following the left and right pattern evenly.

The better the cartridge construction, the better it will translate the movement of the stylus inside the magnetic field it generates. Stylus shape determines how well the shape of the grooves is followed by the stylus. Mostly these are small differences, but differences all the same.

Phono amps need to amplify different tone bands at different levels (e.g. lows need more amplification) and are little marvels in and of themselves. Precision equipment in miniature packages. Again it comes down to physics. Real physics.

If the drive system of a turntable is not consistent, the rotation speed of the platter will vary. This will cause audible variances in sound. Real physics again.

Record cleaning? Well yes: dust and grime will cause unwanted movement of the stylus. This will generate unwanted sound. You guessed it: real physics.

Cost of vinyl? No idea what you are talking about. Cost of physical media has little or nothing to do with sound quality. What has to do with sound quality is the quality of the pressing. Some pressing plants output horrible pressings. Guess what: this is physics again. Bad groove = bad sound.

I’ll give you this though: even when discussing a full analog system, mystification will run rampant. Human nature I suppose.

1 Like

Like I said, and you guys keep proving it. The “quality” differences are full 100% system dependant. There is no such thing as absolute better or worse quality here. There is nothing Roon can do about that, it is 100% up to you to do the right system matching. If you prefer Roon, match your system around it and stop wasting time comparing it to others, it will only lead to dissatisfaction.

2 Likes

This is where your assumption is wrong. You do realize that digital transmission has precisely been invented to circumvent this issue (that every analog stuff will always generate noise and loss etc.)?

There is nothing as noise impacts the sound quality. Your DAC receives the bits it should receive or it does not. And your DAC reads the bits and only the bits or your DAC has a serious issue. The noise « around » doesn’t get read. And if the digital bits are still there, the noise « around » doesn’t change them, that’s precisely the reason why they are digital bits.

You should stop applying some analog principles to digital stuff. Yes the digital bits are transmitted using analog methods of course, but they are read in the digital domain.

This is why we can send images over the Internet thousands miles away with dozens equipments in between.

This is the beauty of digital. Either you have the digital bits or you don’t. If you have even the slightest tiniest bits, you got the information, period. Again this is not an analog signal.

You know, this is exactly where everything is suddenly falling apart. What you hear is impossible. Bear with me, I’m trying to reach your side of the story here.

When the digital bits are impacted, the result is random. Again it’s not analog. When your vinyl is slowing, you hear the music slowing, because the information is encoded as an analog stream and read as an analog stream. When your needle is of better quality you hear better stuff, because the needle translate more of the information present on the medium. When your digital information is impacted, there is no way it is resembling what it was initially. You can’t hear more « airy » stuff or « micro dynamics » or whatever. The information you got is totally random from the initial information.

If you throw a dice on the table, there is absolutely no logical link between the 3 you got the first time and the 5 you got the second time.

0000010 is not close to 0001000. From an analog perspective it may be. From a digital perspective it is absolutely not. If the first one is a sound, the second one is something else. It’s not « voice emerging » it’s not even a sound anymore.

This is why when you have an issue in the digital domain it is very obvious.

This is very important, when the digital bits are impacted, they are no more here, or they are random, there is no more of the initial information, even residual, so what you will hear is dropouts or clicks or random garbage.

This is a digital issue:

This is an analog issue:

Why this is so important? It helps us separate what is trying to fool us from a real issue. It is very easy to hear issues in the analog domain and it is very easy to hear issues in the digital domain. And they are very very different issues.

You should try to convince yourself of the logic behind all this, it has been a great relief for me to finally settle some clear principles in my mind and to prevent him to fool me for more quests instead of more music.

Sorry for the very long comment.

12 Likes

This is a truism since the world at large operates with analog quantities (at least if we’re not talking about quantum effects). The same statement applies to any digital information; an audio signal is not a special case.

We are presumably talkiing about transmitting data by Ethernet or USB, so let’s look at those protocols in more detail.

If you focus on the time at which an individual bit is recognised by the physical layer (PHY) device as a zero or one, it is undoubtedly true that the timing may be affected by noise in the analog, “on the wire” representation of the digital bitstream. However, Ethernet and USB are deliberately engineered so that these timing variations do not compromise the end-to-end integrity of digital data transmission. (For example, this is one reason why there are maximum allowed lengths for Ethernet and USB cables.) That’s to say there is a timing window within which the bit transition needs to take place in the receiver’s PHY for the resulting zero or one to correctly reflect the transmitted data. If the transition happens within that window, the bit has the correct value. The PHY also recovers the effective bit clock from the incoming on-the-wire signal. Just as the voltage levels representing zero and one are “quantized” by reference to a given voltage level, so the timing of the bit transitions are “quantized” by reference to a reference timing signal. The PHY thereby produces a stream of data bits at the effective bit clock frequency, where variations in the incoming analog, on-the-wire signal are ironed out both in terms of data value and clock timing.

It’s possible that some noise or interference might affect the on-the-wire signal and upset the receiver’s PHY badly enough that it recovers incorrect data or fails to regenerate a stable clock signal This could result in bit errors in the data stream being clocked out of the PHY to the Ethernet MAC (for example). This should happen only very rarely, if the system has been built according to the relevant standards.

Both Ethernet and USB are packet-based protocols which have built-in error detection mechanisms at the hardware level; these use CRC algorithms over the contents of the packet. If a large-enough timing variation – or some other noise or interference – results in the contents of the packet being corrupted then the hardware will generate an invalid (non-matching) CRC and will drop the packet. Such packets affected by noise or timing variation in the on-the-wire signal will normally never reach the software layers of a music renderer.

Since Ethernet and USB send data in packets, the audio samples are necessarily transmitted in bunches, so N samples (with N depending on the bit depth and audio sample rate) will arrive in the receiver (e.g. DAC) at essentially the same time. (In pretty much all modern hardware designs, the Ethernet or USB controller hardware uses a direct memory access (DMA) mechanism to store valid incoming packets into memory; consecutive audio samples will be put in memory nearly instantaneously.) Because, in the short term at least, the samples don’t arrive at the audio sample rate, they have to be buffered in memory in the DAC. The DAC will use an internal clock to clock samples out of the buffer at the chosen audio sample rate; this internal clock should be completely independent of (asynchronous to) the Ethernet or USB clock. Therefore, timing variations (jitter) in the Ethernet or USB clock recovered by the PHY do not change the rate at which audio samples are clocked out of the buffer and converted to analog; that process is determined entirely by the DAC’s internal clock. With good reason, the designers of high-quality DACs usually go to some lengths to ensure that the audio sample clock has high precision and stability and is not unintentionally “coupled” to the relatively noisy Ethernet or USB clock.

If there is still an engineering problem to be solved here, as you assert, I think these questions need to be answered:

  • What precise mechanism(s) allow noise and/or timing variations in the Ethernet or USB wire signal to persist through the PHY, Ethernet or USB controller, and buffer memory of the DAC to affect its analog output?

  • What is the relative magnitude of those effects on the audio signal at the output of the DAC?

9 Likes

The way these replies are getting longer and longer has really proved that there’s 10 people in the world - those who understand digital information transmission, and those that don’t.

5 Likes

I’m hesitant to wade in, but it seems to me that:

  1. The fact that digital signals are transmitted as changes in voltage levels in an analog domain (eg, sometimes they hit 4.9675 or 5.0321 instead of the 5.00000 that “means” a 1 vs. a 0 doesn’t inherently mean the signal is analog; it just means that it’s a well engineered network transport layer that’s resilient to background noise. If that wasn’t true, none of your network dependent synchronous tasks would work - you couldn’t browse the web, nor log in to online banking. Sometimes you have network layer glitches, and as @thumb5 was saying, you get time-outs or slow-loads. But that is the network layer saying “hey software, you gotta hang on, I’m seeing some problems here, I’ll be back once I have a clean data stream to present you”. You can experience that as a spinning cursor, but have you ever seen a web page display gibberish or a garbage image? I personally have not - not even a single typo or pixel out of place. Just delays. Which you would experience when transmitting an audio file as the music stopping - the network layer would say “I can’t keep up with this error rate, so I’m not going to present a data stream”. It’s true that in video streaming, you can see pixelation while a video stream is buffering - the makers of the codecs and the display layer have decided that they’d rather present you a degraded stream faster than waiting to present you a perfect stream, because they know that if they lose your attention they lose it to a competitor. But that implementation of falling back to lower resolution doesn’t exist in FLAC/MQA/audio as far as I know anywhere - because the data streams are so much smaller than video that it’s not necessary, and it’s not entirely clear what the “degraded” experience would be like.
  2. The fact that electrical noise can pass down any cable (air is a better insulator than metal) is true. Electrical noise can enter a player from a streamer. That’s why many people prefer SPDIF to USB, or especially AES (which goes over a grounded cable); TOSLINK really cuts any electrical connection but comes with other trade-offs. But there are plenty of good implementations of USB that isolate the receiving and reclocking electrically from the DAC portion of the board. Unison by Schiit is one that has made a lot of impressions and is very transparent about what they’re doing, but there are plenty of other good ones, many by very talented engineers. But the basic idea is that “yes, electrical noise can pass along a cable, so if you think you can hear it, there are ways to minimize or eliminate it”. I personally use WiFi on a couple of my Ropieee’s into SPDIF into my DAC, not for this reason, but because it’s convenient. I’m very confident that electrical noise is not passing into the Ropieee (air-gap), and I’m highly confident that it’s not passing over SPDIF. At some point I’ll try over AES into my one DAC that will support it. Some people at SBAF say that is even better. I’m skeptical, but open-minded.

I think that this unfortunately is one of those topics where some people say “it doesn’t matter what you think should be possible, I know what I hear”, and others say “it doesn’t matter what you think you hear, I know what is possible”. I unfortunately have some sympathy for both sides.

There have been some changes in my system that “shouldn’t work in theory but do”; I tend to keep them because I like them. There have been some that “should work in theory but don’t”. I tend to try to be rooted in my understanding of the science, but at the end of the day I’m constrained by my budget, my time, my other interests, and my willingness to engage in shouting matches. I personally think that Roon does a good job passing bit-perfect streams to my DACs, and any variation in how it does so vs. other streaming software to the same transports is interesting to me, but so far below the level that I can make by swapping speakers/headphones or even DACs that I’m personally not able to make the time to engage. YMMV as with all things, as may your feelings about it. More power to those who engage in various systems tweaks that I don’t spend time or money on.

Still, I am guessing that Roon, as with any other company, is assessing the market opportunity of doing lots of things, and has to figure out which features/processes will address which segments of the market. My intuitive feeling is that there are more folks and $ broadly in the “I like the interface, and the SQ is the near-enough equivalent of bit-perfection and I can control sound at the transport layer on down” vs. the “I will do everything in my power to squeeze SQ out of the streaming layer” camp. But I can’t know that without doing market research, and neither can any of us - reading forums is a notoriously poor way to assess that (because I am confident that the vast vast vast majority of Roon subscribers and potential customers are not on here, Head-Fi, SBAF, Audio Nervosa, or anywhere else findable with a google search).

Best of luck to those on both sides of this discussion!

5 Likes

I don’t think I know what any of that means, let’s alone all of it. But it seems you have a streamer, an Innuos Zenith, that sounds worse to you when you use Roon compared to Squeezelite. This is really an issue you should be taking up with Innuos, isn’t it? Or discussing with whoever designed your DAC? Expecting Roon to “fix” something that is an issue with another manufacturers hardware is a bit much. At least I know not to buy an Innuos! Aren’t they the company that sell you a £3k box to convert USB that comes out of one of their products into USB?

3 Likes

Yes the Phoenix reclocker. 3k to add to a 3k server.

£3k to turn USB into USB! A bargain! Would two be even better?

1 Like

Everybody’s free to do whatever they want with their systems and even enjoy the outcome, no matter the theories, physics, or technicalities behind all that.

But, while it’s entertaining to read about it to some degree, I just wish that combatants would refrain from trying to defend their points, using scientific and technical terms without having proper understanding of them - I’m so tired of alternative facts!

1 Like

That one is awesome, yes its expensive but it makes a big difference. Having said that, I can’t explain why USB protocol clock matters (the 24MHz one) but you don’t need golden ears or a very transparent system to hear what it does.

1 Like

But if it DOES something, shouldn’t the maker be able to explain and prove with measurments exactly what it does?