As a long term Asset uPnP user, having recently aquired a new NAS (QNAP HS-251+) I decided to play about with Minimserver as an alternative.
I need to do some more detailed listening, but my first impressions are that there is a quite significant difference in sound quality between the same files streamed via Minimserver and Asset - and this is just listening through a 1st generation Muso!
(My second impression is the pernickitiness of Minimserver in the files it doesn’t like - your metadata and file types needs to be super-strict but I’ve actually only had a dozen or so errors to fix which isn’t bad out of my total nearly 2,000 albums. Also I’ve yet to find out how, if possible, I can edit the indexTags property but that’s another story!)
Has anyone else noticed a difference? (I won’t tell you which one I prefer yet!)
In theory there should be no difference in sound between any media serve software. It only facilitates a data stream connection between server & renderer, it does not carry any data as such.
I’ve tried a few on my Synology, their native ‘Media Server’, Minimserver, Plex & finally Asset UPnP when that became available for Synology.
I could not detect any change in sound signature that was that noticeable other than maybe in my imagination between any of them when playing straight WAV, although I only compared Asset with Media Server & with that I felt Asset had a more natural sound…
Playing FLAC was the same, nothing much different in sound that I could detect.
I did hear differences with transcoding from FLAC & also noted differences in operation; however it was not something I’m concerned with as all my files are WAV or DSD.
Media Server was easy to set up but had little to offer with customisation.
Minimserver was a bit of a faff to set up (in comparison to others) in that it needed Java & Minimwatch in addition
Plex was OK but it’s designed primarily for video & takes up a lot of space & CPU (in comparison)
Asset is easy, its a stand alone package (doesn’t need extra software)
Interesting, I have a Melco N1a, USB connection to an Audiolab MDac+. I installed Bubble upnp on a Samsung tablet. A conversation made me want to try Linn Kazoo but it would not work at all from the tablet. So I installed it on a Huawei phone. I have been convinced that it is subtly, hard to define a better listen. I don’t use it because it is inconsistent, will not always play an album all the way through.
I had thought of posting about it, but forgot until reading this thread.
Hi Catswhiskers, I’ve tried Bubble & Kazoo & as good as they are they are not a UPnP/DLNA media server software, they’re apps to access whatever the thread subject UPnP media server software is installed on the server (NAS, PC etc)
I think you are onto something here - I have a Netgear ReadyNas Duo v1 with Ready DLNA enabled, streaming uPnP via ReadyDLNA sounds significantly worse than streaming the same file using the uPnP server built into the HDX/NS0*/UnitiServe via a Naim streamer (272/NDX etc).
That’s a completely different situation from the one discussed by the OP, though.
I understand that the OP is finding that Asset and MinimServer sound differenty when they run on the same computing platform and with the same network and renderer setup.
That the same UPnP server can sounds slightly differently on different computing platform is a fact that has been investigated and reported in this forum, by the way.
Thus, it is also conceivable that, on a given platform, Asset sounds better than MinimServer and, on another computing platform, the other way round.
I would expect such differences to be negligible, however. Thus, I find the findings reported by the OP a bit surprising and therefore interesting.
Yes absolutely… I undertook quite some analysis a few years ago and I determined the causes that resulted in difference in SQ on UPnP playback.
I shared my findings with Naim whilst they were developing the newer streamers. The streaming spooling method in the new devices is now totally different, and the differences are not apparent to me now.
My favourite sounding UPnP media server was rather surprising… and in measurement did have an indeed rather specific signature.
BTW the differences are NOT specifically caused by media server software itself such as Asset, Mi imserver etc…
In summary I found a correlation between the inter frame timing consistency on the ethernet and the sound signature on the legacy streamer. The more consistent the spacing between the frames, during the media transfer, the more I seem to prefer the sound. The greater the variation the less i found the sound appealing.
This led to experiment further and I discovered the benefit of the internet streaming proxy media server with services like Tidal - it seemed to vey much improve the sound as it made the inter frame timimg look and sound like it was being locally streamed.
The timing consistency seemed to be a product of the hardware, operating system and to some extent the application working together.
The most consistent timing I found was from a Netgear RN102 running its internal DLNA server - and that sounded the most preferable on my system over the other platforms and other applications.
I created some wireshark macros to explore this further. Look at the red line which is the minimum time between frames in media transfer bursts
Hi Simon, Could you make these macros available, as I would be interested in measuring this, in particular Asset on a RPi2 with a very leanOS and the ‘bridged’ UPnP output from the SonoreUPnP bridge product feeding my NDS/555DR from a ROCK based NUC.
Thanks
You ideally should configure and use a SPAN port on a managed switch such as a 2960 or similar that is monitoring the switch port to your streamer and feed the SPAN data into a dedicated monitor interface with no other traffic. I use a separate USB ethernet interface on my iMac which I configure Wireshark to monitor. That way you are are relatively close as you can be to the actual frame timings
Hmm - I wouldn’t normally post such things here - but Mr Pepper did request - so excuse the rabbit hole…
But in summary: The best SQ was attributed (by a few listeners inc myself) to the media server that had the smallest interval between media frames during a media transfer burst… this method of spooling was changed with the new generation streamers and the effect is now not apparent to me and at the time Naim engineering design were fairly confident my observations would not be so relevant with the new generation streamers. The instrument under test was an NDX.
Indeed, as interesting as it might be as to why, the question is a bit simpler - do they sound different & on/with what.
@nbpf post above sums it up … "I understand that the OP is finding that Asset and MinimServer sound differenty when they run on the same computing platform and with the same network and renderer setup.
That the same UPnP server can sounds slightly differently on different computing platform is a fact that has been investigated and reported in this forum, by the way.
Thus, it is also conceivable that, on a given platform, Asset sounds better than MinimServer and, on another computing platform, the other way round."
Thanks - unfortunately the 5-port switch I use for my ‘front-end’ HiFi componets is an unmanaged Cisco SG100D-05 model, so no method of port mirroring. I do have an 8-port and 5-port Netgear Managed switches in the backend, but that’s a sub-network away in terms of isolating the traffic between streamer and the UPnP server/provider.
So will have some network re-configuring to do, if I want to ‘watch the wire’ going into the streamer.
