So what Upnp do you use and do you use it as a stand alone or you still use it with roon?
IF there are any differences in sound quality between playback software the differences are so small that it is imposiible to say which one is better, it is completely system dependant. Changes that you “think” you hear, changes that you only hear after intensive evaluation, changes that sound better on some music but not on others are are all totally not worth worrying about or at least they are not changes you can expect any software developer to do anything about because they are 100% dependant on YOUR playback system. Roon is bitperfect. If it sounds any different than other bitperfect playback software it is because of a flaw in YOUR playback system, not anything a software developer can fix in a general way. If your playback system is up to par diferences between various bitperfect playback software are non existant in my experience.
4 Likes
@F208Frank when i play UPnP I am using following setup:
minimserver -> BubbleupnpServer (making Open Home proxy) -> DAC
As a client i am using Linn Kazoo which supports only OH (that’s the reason I have to use Bubbleserver). I prefer this client as it’s nice looking and working fine.
OH is just extension for UPnP which allows for gapless playback for e.g.
The differences in my case are on subtle level, however still possible to discern via ABX testing. Roon is bit more grainy , while UPnP is more fluid, airy, realistic.
My personal tip of what is the culpris is:
- RAAT requires more CPU usage on the renderer
- RAAT required additional network traffic which is not present with UPnP (or at least in not such high frequency/volume)
So at the end it’s up to SW and HW implementation of renderer how it can tackle with processing of RAAT stream and I do not mean bitperfect here as RAAT can be easily bitperfect, but i mean jitter and EMI caused by renderer CPU which i believe are the main reasons for this SQ topic…
EDIT: oh i have forgot to mention that I even created my own streamer (RPI3 based) with extreme optimizations for EMI/RFI filtering, galvanic isolation etc. Results are the same as described above with PS Audio Bridge2 streamer. Some photos to get an idea are here: https://localhost.sk/g/index.php?/category/1131
1 Like
I have a Naim Atom and UPnP and Roon sound the same. I have Asset running on QNAP which transcodes flac to wav before sending it on to the Atom and Roons on a dedicated ROCK system. ROCK has locally attached files as NAS is backup, Asset uses the backup. Even with different hardware and software to me on my system they sound the same.
There are cases when software can affect the sound, for example if you connect with USB between the computer with the music player and the DAC. The reason is that the more a CPU works, the more electronic noise is generated. And if graphics hardware is used it adds to the noise.
This is one of the reasons why Roon recommend that Roon core and the endpoint should be on different machines.
I have tested this intensely in the past with cpu stresstests. A IT friend wrote a script for me that randomly put the cpu to max 100%. We listened with three person and none of us could tell a difference between aprox 4% cpu and 100% cpu. One person ‘thought’ he heard a difference but at that time the script wasn’t running. This was with a USB to AES converter on a windows desktop pc. The same test was done with the PC in the audiorack on the same mains outlet. We listened to a analogue source with the computer completely unplugged, with the computer powered on an idle and with the computer at 100% load. None of us could hear a difference.
2 Likes
I think the bottleneck is that bad measuring dac instead of Roon or the raat protocol.
When I had my DAC connected directly to my computer with USB, I got loud ticking noise when playing games, an extreme case I know but it shows what lots of electronic noise can do.
This is very system dependent, but my point is that music playing software can make a little difference, and in this case Roon is pretty bad since its a very “heavy” player using hardware accelerated graphics.
Does this apply to a Intel NUC runing Roon ROCK also the use of hardware acceleration?
I can use Qobuz both in Roon or directly via dCS Mosaic to my Network bridge and I cant say that I can hear any difference between them.
No. There’s no GUI involved in the Roon Core running on ROCK.
That sounds more likely like a IRQ conflict rather than ‘noise’. Games are known to be very resource hungry.
I know that there is no GUI as I use ROCK but my question was more if Roon can use GPU offload like HQ Player can (I suppose they dont do that).
Roon don’t offload anything to GFX card except the OpenGL used for GUI as far as I know. Besides, the typical NUC don’t have any dedicated GFX card. Still, even a NUC with a very slim Linux does produce some noise, which is why a more lightweight endpoint might make a small difference (or so I have heard, haven’t tested myself).
I think something like this is best:
PC with Roon GUI -> ethernet -> NUC with ROCK -> fiber -> high quality streamer -> DAC
Please do. Many people over here are repeating exactly the same as your statements without ever having tested it by themselves. The more people repeat it the more it becomes the ‘truth’.
There seems be only one focus around here and that is ‘noise’. This has been repeated so many times that a lot of people really do believe everything is related to ‘noise’. Well some is but by far not all difference in sound quality is related to noise. Specially audible distortion, hickups, skipping, ticks and pops are not caused by ‘noise’ You certainly can’t fix it by throwing just some linear power supply somewhere in the chain.
I’m sort of reading ‘high quality streamer’ as ‘streamer that doesn’t mess up a flaky DAC’… but I guess that’s not what’s meant in this context…
1 Like
Once you have a bit-perfect stable connection, then there are only 2 things that can affect the sound in a digital audio chain:
- electronic noise
- timing error (jitter), which mostly applies to SPDIF
For asynchronous USB, timing problems are almost non-existent if correctly implemented, so only electronic noise remains.
Here is my setup, which isolates the HiFi setup from the computer:
- power amps run from dedicated fuse
- ethernet switch -> fiber -> fiber media converter -> streamer
- fiber media converter, streamer and DAC run on LiFePo4 batteries (lasts more than a day so charge during night)
- Pin1 on XLR between DAC and amps removed (ground), to make DAC and streamer floating
2 Likes
@Magnus I have quite similar approach as your setup and could easily hear improvement over standard metallic connections to switch.
My further comments:
-
ad XLR pin1 (ground/shield) - generally it’s recommend to have ground connected on one end and disconnected on the other. With this the cable would be still shielded. However not sure if there would be any benefit as XLR is balanced connection and should be differentially cancelling any external noise anyways (might be good “project” to test this during theese days we are supposed to stay at home
) -
ad USB async - actually usb interface (like XMOS) is known to generate noise (i know some people are already allergic to this word, sorry
) at 8kHz , disconnecting USB power line is recommended and also proper cable shielding should be in place (see previous point). Additionally one might consider galvanic isolation for each USB line or eventually even signal regenrators - there is quite a lot of products in this area (singxer, USB Regen etc.) -
ad metallic connections in general - my experience is that disconnecting all metallic connections (TV cable, ISP connection, SPDIF coax connection from that old DVD player used by kids, etc) from audio chain except those really needed improved SQ. I was really pleased when tried this and decided to keep it that way (now using only Toslink optical to connect multimedia to audio chain).
I would think removing pin1 on one side could be beneficial even without using batteries, as long as DAC is grounded through another path (you get rid of a small ground loop).
I have made my own XLR cable, using only the signal cables (pin2 and pin3) in a 4-braid like this, tried with shield but that sounded worse (dampened sound):
2 Likes