Build 99 does it sound as good as the previous build to you?

Roon Software Discussion
21

We’re not out to ridicule anyone. There is definitely a place for subjective observations when evaluating audio playback systems–virtually all well-respected audio manufacturers engage in subjective testing as part of their product development process.

Roon is still passing its bit-perfect tests–so this isn’t that sort of problem as long as you’re seeing the purple signal path light. That leaves us with the analog domain.

Before we go there, keep in mind that Roon does a lot of things that many other apps do not, including but not limited to:

  • Painting a pretty, GPU-accelerated user interface
  • Background analysis of audio files
  • Background processing of metadata updates
  • Background syncing with TIDAL
  • Scanning for changed media files
  • Acting as a server for other control points in your house

All of these things can make your computer work harder than it might otherwise, and if your system is built in such a way that analog-domain noise from your computer has a meaningful impact on your analog audio gear, it is possible for just about anything going on inside of the computer to have an impact on SQ.

If you’re going to make a serious evaluation of analog-domain behavior, it’s necessary to control for what every other app on the machine might be doing, as well as your operating system. And, for that matter, what other people in your house sharing the same circuit as your gear are doing at the time. And I hope that none of the other software on your machine has taken an update recently!

Right when Roon starts up, it does a full sync of your TIDAL library (if logged in), and a full walk of all of your storage devices looking for new/deleted/changed files. Depending on your storage configuration, this can take quite a while.

Even the exact state of the UI can make a difference in your CPU/GPU workload: if the seekbar at the bottom is ticking, the screen is painting the screen ~1x/second. If the nowplaying indicator on an album details screen is animating, it’s painting ~15x/second. If the spinner at top right is spinning, it’s painting ~20x/second.

Regardless, background work, or having a nice user interface, aren’t bugs. It’s just part of what makes Roon Roon. We can’t provide the user experience that we provide without that stuff.

This is part of why we’ve developed RAAT and the RoonReady program–it removes the conflict inherent in juxtaposing “powerful media server that runs on a fast noisy PC” and “analog circuitry critical to sound quality”. I guarantee that we wouldn’t be having this conversation if your audio was flowing through an Auralic Aries or Meridian 808.3 two rooms away from the server. Those devices, unlike your computer, are carefully designed and measured to ensure impeccable analog-domain behavior.

There’s nothing wrong inherently inconclusive about subjective listening tests if they are done in a statistically sound way–blind ABX testing, preferably with at least a small population of listeners. Personally I don’t even trust my own subjective observations if made outside of a structured and controlled test environment. These testing methodologies are sound, and were developed for a good reason.

This is difficult stuff, and a touchy subject for many. We’re happy to help in any way that we can–but there is simply no practical way that we can fix a problem that’s described in terms of how “analog” Roon sounds. We need something more concrete to investigate.

1 Like
22

Brian: What further info would you like me to provide? System components,
listening protocol, file and tracks listened to… At your disposal.
Didn’t intend to touch a nerve. Thanks for the detailed response.
Let’s together investigate further.

23

I want to hear about all of that: your platform/setup, how you’re testing, etc.

And I want to give you some testing guidance, since there are probably a few easy things that you could be taking into account that might help smooth out sources of error in the tests.

Since we’re pretty confident that Roon is bit-perfect, at least up to the interface with the driver, we’re looking for something not directly related to audio.

Here’s a bunch of ideas. Some are things to control for. Some are other troubleshooting techniques.

(1) In settings -> services, look “Syncing library now” + wait for it to finish:

Screen Shot 2016-01-25 at 10.23.59 PM.png693×91 8.08 KB

(2) In settings -> storage, look for “Scanning now…” + wait for it to finish:

(3) In settings -> setup, disable both analysis settings.

(4) The spinner in the upper right covers importing and metadata lookups. Wait for that to finish, too.

(5) Monitor the Log.

After starting Roon, open up terminal.app and run:

tail -f $HOME/Library/Roon/Logs/Roon_log.txt

(Replace both incidences of Roon with RoonServer if you’re using RoonServer)

When Roon is at idle, and you’re not playing music or clicking around, there shouldn’t be tons of text flying by. Roughly every 15 seconds you’ll see one of these:

00:13:16.155 Info: [stats] 3196mb Virtual, 412mb Physical, 54mb Managed, 31 Threads, 120 FDs (73 REG, 25 IPv4, 8 DIR, 5 KQUEUE, 4 CHR, 3 PIPE, 1 systm, 1 unix)

If you see a ton of stuff, then Roon is working–and there’s a chance that this is polluting your tests.

I always monitor logs during listening tests to make sure that if something unexpected starts happening that might impact the sound, I’m aware of it.

(6) Monitor CPU usage

This can be another clue of unexpected stuff happening. When Roon is idle, its CPU usage should be low:

Screen Shot 2016-01-25 at 10.46.38 PM.png843×292 53.7 KB

I just took a quick snap of my machine. To my surprise, Google Chrome is doing quite a lot–if you’re worried about things that Roon might be doing outside of the “bit-perfect playback” domain, you should be just as worried about stuff like this. If I saw this, I would exit Chrome before proceeding.

(7) If you have two GPUs, control for it

My mac has two GPUs. I run gfxCardStatus to keep an eye on which one is being used. I recommend holding that constant when A-B testing. You might even want to force your machine to use one or the other to see which way you prefer.

(8) Take the UI out of the equation OR take the server out of the equation

This is a little bit more involved, but if you have two computers, you can:

  • Run RoonServer on the machine with the audio hardware, and run a Roon as a remote control elsewhere
  • Run RoonServer elsewhere, run a Roon as a remote control on this machine, and play back audio using the private zone

You may find that you prefer one, or both, of those over the stock all-in-one configuration.

The ultimate test would be a blind comparison between B94 and B99, set up by someone other than you, and performed by you. We can figure that out if you want to try it, but I think you may have a few things to explore first.

I find it a lot more appealing to isolate my media server + renderer than to troubleshoot computer noise at this level. It’ll be nice when there are a few more options out there and you can just try that without buying something expensive. If you are sensitive to this sort of interference, I suspect that a setup of that arrangement will sound better to you than anything computer based.

2 Likes
24

Brian

Your latest replies are interesting but If, as you say, nothing was changed sonically, miss the point, at least regarding my situation:

All I did was run the update. NOTHING else was changed in my system when I made my observation. Since then I have done 2 things: reboot all my devices and change my Rsync delay from 50ms to 0.

So why would I need to go and look for something that isn’t there?

25

I think its called a soft landing.

1 Like
26

I appreciate those who are skeptical, but I have been a user of Roon since the early days and subscribed to it in large part because it sounded as good as Amarra. Recently, it does not.
That said, I appreciate the responses from Brian and that fact that one could read my reporting of
decreased SQ as either auditory hallucination or criticism of Roon and their extrordinary efforts in
producing something quite special. My concern is certainly not intended as criticism.
That said, I will look at all I can in my system, as Brian suggested, and report back.
Good listening and best to all.

27

So why would I need to go and look for something that isn’t there?

We have entertained the possibility that we might be wrong, re-run the bit-perfect tests, and reviewed the code changes between the two builds. We know that nothing has changed with audio handling in build itself outside of the scope of AirPlay/DSD->PCM.

So, from our perspective, while there may be an area where we can improve, it is looking very unlikely that it was caused by a change in the code between build 94 and build 99.

If that investigation found something, we would have fixed it. Since it didn’t, the next step is to go back and make sure that some other aspect did not pollute the results. Maybe the situation is “Roon sounds bad to me while it’s scanning for file changes” and not “Roon sounds bad to me when running build 99”. My above post is pointed at taking that next step and picking apart those distinctions.

If you don’t intend to cooperate with us in investigating this, that’s your choice. It means that we won’t get anymore information about the problem and the issue is more likely to languish. This is purely a practical matter: we need to understand a problem before we can do anything about it.

3 Likes
28

I’ve been an audiophool for many years and have very much learned to question my “logic” when attempting to ascertain when/why/how something sounds worse or better. Far, far too many variables. Power variability,EMI & RFI are often culprits as is wireless when used along the chain. But, in my opinion, so too is what’s going on in my brain while listening (i.e. that thing that is separating my two input devices ;-).

With that said, if the perceived sonic degradation is consistent day-to-day and switching back to the prior build restores sound quality, it’s not a leap to believe there’s a correlation. But that’s not the same as knowing what caused it.

@brian - I applaud your willingness to help identify potential causality.

1 Like
29

I do wonder what what happen if Roon released three builds at the same time that were identical in all but the build number (but didn’t let on), then asked people to vote on which sounded best.

My own guess is that at least a portion of respondents would enthusiastically rank them in terms of sound quality. :joy:

30

I do think we all know there are things that are not measurable (currently) that do affect sound.
However a thread suggesting sound has dis-improved also affects sound!

It’s actually very difficult (for me anyway) to remember a sound. Sometimes my I think my CD player doesn’t sound as good as it did the day before, then I have a WOW moment.

I think Roon need owe it to users to take this seriously (which they appear to be doing) and certainly they don’t want an inferior sounding product out there. An in-house ABX is possibly called for?

SJB

31

Chrome has stuff running even if the browser itself is not running. That was the reason I removed it from my audio machines because it often decided to start installing updates at inconvenient moments causing drop-outs in situations where I was at 90% CPU load already playing music. Firefox also has background update agent, but that one is easier to disable without harming the browser (you can opt-out from that also while installing).

For a lean good browser I recommend using Opera. Nowadays it uses same Blink engine as Chrome (fork of WebKit).

1 Like
32

Perception is reality, but it doesn’t mean we don’t know what to measure or that it can’t be measured.

33

What I’m trying to say was more eloquently said in Hamlet,

There is more in heaven and earth Horatio than is dreamt of in your philosophy,

You cannot look at just a set of measurements and know that something will sound good. Sometimes something unexpected enhances or degrades sound quality without necessarily being measurable.

It is possible that the measurements remain the same and the sound changes.
It is equally possible that the measurements stay the same as does the sound.

A good blind ABX listening session should sort out which is happening.

I don’t personally notice any difference.

SJB

34

Brian,

I followed your suggestions and am now able to document results: I followed and viewed what was
suggested- numbers 1 through 6. I don’t have two GPU’s or two computers. The only settings I changed in Roon was to switch Background Audio Analysis… and On-Demand Audio Analysis from “normal” to “off” (where they will remain). I do not use volume leveling.
Upon booting the computer (I always reboot before using Roon), I waited for sync and scanning to be completed. Other than Roon, the only program I open is Dirac Stereo- have used this duo since starting my subscription.

CPU USAGE:
SYSTEM: 1.9 to 7.2 (min/max) typical 2 and 1 when Roon at idle
USER: 2. TO 9 (min/max) typical 2.5 and 1 when Roon at idle
IDLE:: 83 to 95 (min/max) typical 95 and 98 when Roon at idle

LOG MONITORING:
At Roon idle, text as expected every 15 sec. Nothing more, other than a brief sync, taking 334 ms.

MY SYSTEM:
MacBook Pro 13in, mid 2012
OS X Yosemite 10.10.3
Processor: 2.5 GHz Intel Core 5
Memory: 4 GB 1600 MHz DDR3
Graphics: Intel HD Gr 4000 1025 MB
Hard Drive: 500 GB SATA DISK

For music, I utilize USB output, two AudioQuest Jitterbugs (one in each USB port), Wireworld Starlight USB cable.

BOTTOM LINE RESULTS:

After booting as usual, changing only the Analysis settings, and using Roon as i typically do-
it sounded great again: The depth, weight, clarity, soundstage was back. No longer was it thin, bright and fatiguing, like is was yesterday.
So…what happened? Beats the hell out of me! Maybe the analysis settings made the difference. Perhaps my computer was doing something weird yesterday (before I was actively monitoring the Log and CPU usage). Maybe there was syncing and/or scanning going on yesterday. Or something else.

The points are:

  1. Roon sounds great again, and;
  2. Thanks to Brian, I now have some diagnostic tools to utilize should the sound degrade in the future.

Thanks again to the Roon folks and Brian in particular for listening, not judging, and providing useful tools.

Long Live Roon

5 Likes
35

Yes, but this is a very different statement than [quote=“Sloop_John_B, post:30, topic:7268”]
I do think we all know there are things that are not measurable (currently) that do affect sound.
[/quote]
Which I do not agree with.

1 Like
36

Awesome posts @brian. Thanks for giving us a great set of instructions on troubleshooting and for your patience.

37

I had intended them to mean the same thing :grin:

SJB

38

Please tell me how you measure digital jitter and digital/analog grain that is (most likely) inherent in your setup to a grade that you can make judges if something sounds good or bad, eager to learn.

39

I don’t think I posted what you think I posted.

40

Crummwoody, what else did you change since going back to Amarra. Any change can affect the whole balance of playback. I had a server with Win 7 that did not sound good running Roon on 2 machines configured as Roon Server. It was too thick and off to me. I changed the operating system to Server 2016 and now 2 machines with Roon Server sounds more fleshed out and real. That was just a software change. Its all part of tuning to your taste.