A few people on the old CA forum posting in that ridiculous thread where everything done makes a “huge” improvement in sound quality hardly qualifies as “a solid case”. In fact, “gains” claimed in that thread should be treated with great skepticism.
3 Likes
(to my ears) Vote 1 right here for the experimental SqueezeLite RAM memory playback settings on ZENith being better than the ZENith as a normal Roon core, which is itself better than the ZENith being a networked endpoint. And just to give you some fodder to pick holes, this is all directly connected to a DAC via usb (all of which are better than the networked version of each).
That said (as said before) these all are inferior to a CD directly in the DAC (Oppo 205). Measured? No. Lots of relative assessment with mine and my GF’s ears blind and not blind? Yes.
And so back to, at least, my reasons for being pro this thread. I hope/expect >50% of my annual subscription is going towards work devoted purely to SQ improvements. And i would expect a CD player to be always in the mix for relative assessment of any software or hardware testing. It’s my relative champ and expect it to be near, if not at, the top of most other people’s reference points.
Personally I think SQ is best when the core is connected directly to the DAC via unshielded USB, disks are thrashing because all FLAC files are being tested for integrity whilst other tracks are playing and CPUs are maxed out because upsampling is enabled and the same PC is busy transcoding a BD rip to H.264. I’ve tried all combinations from decoupled core and endpoint with both running stripped down OS’ and powered by LPSU to my current config which was arrived at by subjectively listening, and this is what sounds best to me. This config leaves my CDP in the dump. I expect Roon to invest 50% of revenue into exploration of further SQ enhancements borne out of maximising hardware workload. Threads like this are important to anyone serious about SQ.
4 Likes
Yes, that’s a very good point. The real degradation probably happens when CPU load is variable, the poor old cpu doesn’t know whether it is coming or going. Roon developers should immediately drop all other work and concentrate on making Roon use 100% cpu load on all cores at all times. Then they can move on to saturating the network and making the hard drives dance.
Hahaha! You’re a few days late for April Fool’s Day! 
I would say, come round here, listen a realise Roon don’t have to do anything else for sound quality.
2 Likes
I certainly hope this whole post is sarcasm.
Roon Core direct to a DAC is the worst I have ever heard Roon sound. Especially if a laptop was the Roon Core. Conversely, using a dedicated endpoint, like an ultraRendu, with a Roon has always sounded the best.
50% of Roon income spent on sound quality? No way. That would not be smart on Roon’s part.
Pretty sure it’s sarcasm.
1 Like
Is that Roons problem, or your DACs? I would argue that it is your DACs problem. What DAC is it that is so sensitive?
It’s not Roon’s problem. It is a hardware/software noise problem. I don’t know of any DAC that is immune to noise on the USB input, regardless as to what might be claimed.
1 Like
That’s why separating the Core/PC form the DAC/endpoint via Ethernet is by far the best way to go, IMO.
I can put my DAC up to maximum volume and place my ear up against my tweeter and I can’t hear any noise.
Wow. I guess I am wasting my time but that is not the kind of noise I am talking about. Maybe you should do some self-edification BEFORE you comment on these technical matters.
4 Likes
Your comment speaks more of a noisy source than the merits of network vs direct connection? Have you tried a low noise core direct to a DAC? For me better than network connected. Though obviously for noisy sources do indeed put an ethernet switch between it and the DAC for some isolation.
@Martin_Kelly If you have a noisy source. If the source is not noisy (e.g. a dedicated server (innuos, antipodes, etc) then the whole network between source and endpoint better be lower noise otherwise it’s not going to do better than the source.
@Jez that’s why it is Roon’s problem as they centrally located in the debate due to designing towards, and advising of, core and endpoint separation over the network. Why would you want Roon to shut the door on any other signal chain? Home networking is one of the biggest IT headaches for most homes these days. So why put all your money on it as the backbone of your expensive sound system?
And @Chrislayeruk really? They’ve hit the end of the road on SQ and no improvements will ever be made again? Try a CD. Most storage based systems aren’t even better than 10 year old CD spinners yet.
General purpose computers are generally quite noisy from an electrical perspective. Simplified special purpose computers designed to be audio endpoints are generally very quiet electrically. The functions the Core is required to do means it will be inherently more noisy than a networked special purpose endpoint.
So, no, my comments do not speak more of a noisy source. They speak to the reason a Core is more noisy than a special purpose endpoint.
1 Like
Would you preface that statement with:
- It stands to reason that…?
- In my measurements of the output from the DAC I have seen that…?
- In my listening experience I have noticed that…?
No, I would not preface it with any of those statements. It is patently obvious that the Core has to be more powerful and have more I/O features than an endpoint. Those power and features mean that a Core will make more noise than an endpoint built with that specific purpose in mind.
Not contesting that a purpose designed end point will be less noisy than a purpose designed server but you have to add to it the crud from the laptop + switch + cable rfi/emf before it gets to the endpoint. All that vs just the single server. As a reference point, one design philosophy of the Antipodes servers is that keeping the signal clean from start to finish is much more preferable than cleaning it up at the end.
What noise do you think makes it over UTP Ethernet? UTP Ethernet uses transformers to isolate the signal. UTP Ethernet will pass on some clock phase noise, but it is great at eliminating noise from power and almost all noise, save some clock phase noise, from the Core. RFI/EMI is rarely an issue with UTP Ethernet as there are transformers at both ends so that noise is removed.
I have listened to several Antipodes setups and continue to prefer the sonicTransporter/ultraRendu combination. And the price is ridiculous for the Antipodes products.