Music server VS NAS for sound quality

ChrisSU posted:

Tallan posted:

frankly I don't trust Naim's backup scheme, such as it is: if the US drive goes kaput the whole thing has to go back to Naim where they will "try to" recover data.  

You may have to return a Naim server to the factory for repair, and yes, that is likely to be expensive, but if you have set up the automated backup built into it, you are not dependent on Naim to try to recover the data. A repaired unit can have the data restored to it from your backup, and while you are awaiting the repair, you can still play your music by pointing a server at your backup folder. 

Exactly. How is Naims external backup any different really? Naim doesn't use RAID but RAID is not backup and only the foolish think that's what RAID is for. RAID is for redundancy not resiliency. In other words, to assure uninterrupted uptime. 

RAID at home is mainly useful for providing higher read concurrency where you have many users of the same NAS or need the NAS for work and may need to work through a disk failure and replace a disk later. If you have just 2 disks and have to choose between RAID1, or no RAID but 1 disk + 1 removable backup, the latter is the only sensible option.

When RAID controllers die you often lose everything without a spare controller of exactly the same model number. And if it was hardware encrypted RAID with the key stored in the controller, even a spare controller won't save you.

For those reasons, a Uniti Core with just one optional drive bay seems perfectly acceptable.

Also bear in mind simple is best. The more complex the arrangement of NAS, RAID, OS and media server software, the more thibgs there are to go wrong and cause tears before bedtime. I think a NAS with a media server is great if you already have one and are very comfortable with it's management. Adding Asset or Minim to the mix is low hanging fruit. But if I was starting from zero, had no NAS, only needed one for music streaming; then I'd absolutely get one of the bespoke solutions. I say that as someone who supports a storage appliance for a living. 

Thanks everyone for your advice! It is not clear cut, but the trend of responses indicates that a server will not sound better than a regular NAS. I am very hesitant to spend money if it will not improve sound quality.

For a similar amount, I could buy an XP5 XS power supply for my ND5 XS. The XP5 XS would certainly provide more of an improvement in sound quality, than a music server used as a NAS. 

Daniel H. posted:

Thanks everyone for your advice! It is not clear cut, but the trend of responses indicates that a server will not sound better than a regular NAS. I am very hesitant to spend money if it will not improve sound quality.

For a similar amount, I could buy an XP5 XS power supply for my ND5 XS. The XP5 XS would certainly provide more of an improvement in sound quality, than a music server used as a NAS. 

I am not that sure if it’s that easy. But at the same time everybody needs to decide what’s fits in their ambitions, budget etc. However besides doing something like buying a ND555 without hearing it, i always let my ears decide and use the forum as a source of information. 

Daniel H. posted:

Thanks everyone for your advice! It is not clear cut, but the trend of responses indicates that a server will not sound better than a regular NAS. I am very hesitant to spend money if it will not improve sound quality.

For a similar amount, I could buy an XP5 XS power supply for my ND5 XS. The XP5 XS would certainly provide more of an improvement in sound quality, than a music server used as a NAS. 

IME, Audirvana+ added to the Mac Mini sets a good benchmark for sound quality.  I run mine through a Meridian Explorer into the Naim DAC 555 — I am not at all sure if the NDX FM sounds any better, though I rely on and love the latter for overall usability experience.

Nick

Harry posted:

I started with the HDX HDD. I compared music ripped by the HDX, internally stored on the HDX, with music ripped by the HDX to a NAS. I could detect any difference. For this reason I ripped to the NAS routinely because I didn't see the point of using the HDX's internal HDD, which if failed, would have to be returned to base to recover my files. Much easier and more scalable to manage my music on a NAS which I could back up, replicate, etc. I then had my HDX converted to SSD. This setup served me well for some years and brought great enjoyment. I consider it to be far enough in front of my CDX2 to allow me to sell the CDP. I never missed it.

When the NDS turned up I kept the HDX as the server because it had always served me so well. And it could hardly interfere with the sound, could it? Well, as it turned out, it could. When the HDX failed and needed to go back to Naim, I hastily installed Minimserver on my NAS in an effort to keep the music flowing. The results stunned me. I'm not exaggerating. The NDX was an audio bottleneck. I don't know why or how. When it came back from the factory, all fixed up and shiny, I asked my dealer to hang onto it to sell. It ended up being traded in against some of the cost of a full SL loom. 

In summary, my personal experience is that there is no difference between HDX internal storage and NAS storage in terms of sound quality/musical enjoyment. When used with an S level dedicated streamer, the HDX acting as a server for NAS stored music got in the way. Removing it and running server software on the NAS improved the sound quality/musical enjoyment. It could be that my HDX was knackered. It's moot. The performance of Minimserver and Asset on a number of QNAPs in my system has been superior, and not by a small margin.

Melco have supposedly swung this back in favour of  the proprietary server model. I've yet to get my ears on one and am in no hurry.

tried melco on loan  ,seems expensive tad better with high rez  . purchase ripnas/zone ripper for convenience . ,revamped with more memory/ 4x2 drives 2 yrs abo.  never failed . forget ive got it.  also run 2nd qnap 210 /asett again runs ,tucked away   no glltches  . sometimes plays  cds  a tad better for my ears

Daniel H. posted:

Thanks everyone for your advice! It is not clear cut, but the trend of responses indicates that a server will not sound better than a regular NAS. I am very hesitant to spend money if it will not improve sound quality.

...

This is a false conclusion: the responses do not indicate tha a server will not sound better than a NAS. Rather, the responses indicate that not every server will sound better than a NAS. This is a trivial result and, I argue, the most that you can expect given your original question.

There are many reasons why people buy devices like those from Melco, Allo, Innuos, HiFiBerry, Antipodes or just use fanless, headless SBCs instead of NAS devices. One is that they have found those devices to sound better in their specific replay setups. Other typical reasons are that dedicated server tend to be simpler, look nicer, make less noise, etc. than NAS devices. Even more important is perhaps the understanding that one does not need a NAS to serve music contents in a household. Indeed, NAS devices are less than ideal for serving music contents: their offer limited support for running UPnP and Roon servers and can be an ostacle to file transfer and backup management. They are great devices, but not necessarily for the purpose of serving music contents.

Thus my answer is: if your plan is to buy an expensive, dedicated music server to put it in the basement or in a cabinet far away from your other audio devices, then buy a NAS or, better, a standard computer. Otherwise ... it depends!  

I've never understood the rationale behind suggesting a better audio quality from these complex 'audiophile' NAS boxes (such as Melco, Inuos, Naim 'Server Core etc.).

By using a type of switch with a high stability physical layer clock, high quality power supply and low RF emissions profile (e.g. switches like the Cisco 'Catalyst' series) one can effectively get the same result electrical result, the only difference being a few μS of additional latency (and compared to the access time of the disk subsystem - including those with SSDs - this is a very minor amount of additional time).  Given the tremendous reliability of these switches there's no problem with buying a secondhand decommissioned unit and then they are very cheap.

I can see the justification for the combined NAS / Ripper systems on the basis of convenience, but it's a lot of money to pay for that convenience.

Huge posted:

I've never understood the rationale behind suggesting a better audio quality from these complex 'audiophile' NAS boxes (such as Melco, Inuos, Naim 'Server Core etc.).

By using a type of switch with a high stability physical layer clock, high quality power supply and low RF emissions profile (e.g. switches like the Cisco 'Catalyst' series) one can effectively get the same result electrical result, the only difference being a few μS of additional latency (and compared to the access time of the disk subsystem - including those with SSDs - this is a very minor amount of additional time).  Given the tremendous reliability of these switches there's no problem with buying a secondhand decommissioned unit and then they are very cheap.

I can see the justification for the combined NAS / Ripper systems on the basis of convenience, but it's a lot of money to pay for that convenience.

One rationale is that of avoiding having to put a switch with a high stability physical layer clock (and it high quality power supply) between the server and the renderer or streamer: devices like the Melco offer a high quality direct ethernet connection to a streamer without having to bother about switches. Another rationale is that dedicated music server typically offer high quality USB and/or S/PDIF outputs for a direct connection to DACs. Many users prefer to built their systems around pure DACs rather then streamers. For these users, a server would ideally have (or be upgradeable with) high quality USB and S/PDIF connection. NAS devices do not offer high quality USB or S/PDIF outputs. Many users find NAS devices inconvenient for serving music, especially when multi-room replay is not needed. Finally, notice that dedicated music servers do not necessarily need to be very expensive: a Raspberry Pi 3B+ with an Allo DigiOne hat is a very competent UPnP server + S/PDIF renderere and a very straightforward way of accessing Tidal, Qobuz and other internet streaming services. For about 150 EUR it represents a great value for money. I am not saying that one should blindly invest in expensive music servers. I myself would not buy a music server with integrated ripping station or one depending on proprietary, poorely supported OS. But I also never managed to find good reasons to buy a NAS for my system!       

OK, yes I get the USB and S/Pdif argument for DACs (well except for the fact that you can take a 'normal' USB output from, say a Mac Mini or a standard NAS and pass it through something like a 'Gustard' to filter it and reclock it to high quality for a lot lower price than the 'audiophile' NAS devices).

In terms of complexity, even Melco's system diagrams illustrate a switch in use (and also a Wireless router, and an iOS or Android device as a controller) so these complications aren't eliminated at all (and the high quality switch is easy: just pick a Cisco 'Catalyst' switch).

I know that setting up and building your own NAS using a Raspberry Pi is achievable, as is putting a reasonable quality NIC and possibly reasonable quality audio interface onto it, but what you've done is still to build a custom NAS system!

Yes I get the convenience factor, but I still don't see what makes these 'audiophile' NAS devices so special as to justify their prices (magic beans perhaps?).

I find it interesting that on other discussions forums it is being accepted (not just "argued") that a server may have more effect on sound quality than the DAC itself. Cannot reconcile those views with the prevailing one on this thread (apart from the fact, already pointed out above, that "over there" USB is being predominantly employed rather than SDPIF...

DrPo posted:

I find it interesting that on other discussions forums it is being accepted (not just "argued") that a server may have more effect on sound quality than the DAC itself. 

Yes but after spending any time on other forums, I often feel my IQ has dropped 50 points. 

We have some cherished eccentrics here and a cross section of knowledge and opinion. I tend to find the other forums I come upon (like the unsavory coloured fish) to be populated disproportionately by knuckle draggers that can be made to believe in anything so long as it is nonsense with little or no grounds in science or technology but reject anything based on basic fundamentals of engineering and physics. There will always be those that say "The science says it should make no difference but I know what I heard and it was better". Which really means the person:

a) Is admitting to a belief in magic.

b) Is hearing the price tag.

c) Is hearing an improvement and doesn't actually understand the science of it.

I bite my tongue and let others have their opinion but feel there are a lot of As, half as many Bs, and half as many again of type C on most forums.

But the same could be said of this hobby in general. Whole product concepts costing the earth have made it to market based on incorrect understanding of unsound principles or even pure science fiction, and yet found buyers.

No one ever lost money underestimating Joe Public's intelligence but they sure have lost money by overestimating it.

That said, I think the Core is the right form factor, simplicity, and cost package for the customer base that is either knowledge/time challenged. If someone thinks a Melco at twice the cost sounds better than a properly set up NAS with a stable network that's up to them. Just like any component, people are free to experiment and reach their own conclusions which others may find barmy. 

I will say this, if there was absolute truth in this business, then there would only be one hifi manufacturer on earth and everyone would own the exact same units in their budget band and we'd all be bored as hell except for maybe different speakers because our ears/rooms differ.

FZ, nice summary.

BTW, I sometimes fit into category 'C', sometimes because of known unknowns (e.g. I know I don't know enough about the physics or biology underlying the effect), sometimes because of unknown unknowns (and I don't know what they are! ).

That's why so many of my posts are guarded with "I believe that..." or "maybe" or "may" or "I found that..." or just "or..."; these equivocal statements may seem irritating to some, but there is a reason for them.

DrPo posted:

I find it interesting that on other discussions forums it is being accepted (not just "argued") that a server may have more effect on sound quality than the DAC itself. Cannot reconcile those views with the prevailing one on this thread (apart from the fact, already pointed out above, that "over there" USB is being predominantly employed rather than SDPIF...

I have been following the monster thread "A novel way to massively improve ..." on the CA forum and overthere some folks have indeed argued that improving the quality of USB and/or S/PDIF outputs brings more sound quality per dollar than improving the quality of DACs. This, however, seems to be a standard "source first" argument widely accepted also in this forum.

When we compare judgements made here to judgments made over there, we should alo probably consider the fact that many Naim users adopt solutions in which DAC and renderer are integrated in the same box: a streamer. Conversely, at CA, many users tend to rely on standalone DACs. These are typically fed by USB or S/PDIF streams (in contrast to TCP/IP data packages), hence the focus on the quality of the USB and S/PDIF outputs of servers and network players. The fact that most DACs have USB inputs explains why the quality of USB outputs is more discussed than the quality of S/PDIF outputs.

Thus, in a nutshell, I do not see obvious contradictions between the wisdoms that prevail here and those from over there, just different focuses and concerns. These can be explained by the fact that the folks over there tend to use standalone DACs whereas the people here tend to use streamers.

Huge posted:

1) OK, yes I get the USB and S/Pdif argument for DACs (well except for the fact that you can take a 'normal' USB output from, say a Mac Mini or a standard NAS and pass it through something like a 'Gustard' to filter it and reclock it to high quality for a lot lower price than the 'audiophile' NAS devices).

2) In terms of complexity, even Melco's system diagrams illustrate a switch in use (and also a Wireless router, and an iOS or Android device as a controller) so these complications aren't eliminated at all (and the high quality switch is easy: just pick a Cisco 'Catalyst' switch).

3) I know that setting up and building your own NAS using a Raspberry Pi is achievable, as is putting a reasonable quality NIC and possibly reasonable quality audio interface onto it, but what you've done is still to build a custom NAS system!

Yes I get the convenience factor, but I still don't see what makes these 'audiophile' NAS devices so special as to justify their prices (magic beans perhaps?).

Huge, I have already answered your comments but, for some reasons, my post has not yet been approved or rejected. I basically agree with your observations but let me point out that:

1) Some folks at CA seem to have made careful comparisons and found out that, even if one inserts very good isolators between the USB outputs of servers and the USB (or S/PDIF) inputs of DACs, the impact of clocking and power supply in the server on the quality of the signal that comes out of the DAC is significant. If these observations are accepted, investing in a (USB and/or S/PDIF) server that provides high quality outputs can make sense. Of course, it also begs the question of whether a streaming based solution wouldn't in the end be more cost effective. Here Naim have a very strong argument (you can get very high quality outputs from a streamer without having to invest too much money in a server) but also weaknesses (you need a wired connection). I personally think that there is plenty of space for both approaches but I certainly would not put extra money on a server if this had to be used just as a NAS!

2) If the Melco operates as a server from local storage, a LAN connection is only needed to control the device. In this setup, a NAS and a high quality switch are superfluous and one can just use any switch or a direct connection to the router, I understand. Of course, a streamer has to be directly connected to the Melco and the Melco has to act as an access point for the streamer. In other words, the complexity of a high quality switch with clean power supply are hidden inside the Melco. This justifies (a small part of) the extra price tag.

3) Of course I have built a custom NAS system. But it is one that also runs a renderer, has no noisy fans, no arrays of disks, high quality S/PDIF output and very small footprint! I also do not have to control my NAS through an ugly web interface!

What I take from this thread is that the differences in SQ to the extent that they do exist are too small to worry about. 

Being a cost conscious person I’ve tried the NAS route, but got fed up with being a part-time network administrator instead of music listener and bit the cost bullet on a UnitiServe special offer just before the Core was announced. 

My user experience (spdif into dac on my main system, tcp/ip to my streamers) has been dramatically improved and I can do everything I want to from the Naim & Nserve apps on my iPad. For me the UnitiServe has been one of the best hi-fi “investments” I have made and there is no way I’d revert to using a NAS. 

As for VFM, I choose to compare it with an iphone x (from £999) -which I don’t have - and thus regard it as very good VFM. 

Olly

I actually use UnitiServe as my ripper and UPnP server. NAS is a music repository only.

Prefer it this ways - sound quality and ease of use wise.