When overriding the default plays with settings and turning off UGPM It would appear I never get a burst of white noise on about 5% of 24/192 tracks and a similar frequency of 24/192 tracks stuttering. I have however observed my old friend….. the failure to record the last track in a queue as played on about 20% of play queues (same issue I have observed with KEF speakers and my earlier Devialet Phantom II’s for the last few years) with UGPM on this does not occur.
options therefore are
leave UGPM on with max sample rate of 96 kHz which results in no stuttering but rare bursts of white noise on 192khz files (not speaker damaging but terrifies the wife’s cats and could induce a heart attack in me)
turn UGPM off and have to replay final tracks in queues to MacBook speakers with the sound off to ensure accurate play counts/history.
Or option 3) switch to an alternative non UPnP playback software which does not suffer this issue with any of my endpoints.
To be quite honest tired of messing, just want to playback local music files to streaming endpoints with no issues, prepared to accept a minor loss of SQ in return for a no hassle experience…….
Why not just limit the playback sample-rate to 96khz max and forget about it… Up-sample all PCM using ‘Power of Two’ strategy to produce the logically correct sample-rate, so the 44.1kHz family of sample-rates play as 32/88.2kHz and the 48kHz family of sample-rates play as 32/96kHz… Any file above 88.2kHz and 96kHz will get down-converted to the logically correct sample-rate. Convert any DSD files to PCM and they will play at the logically correct PCM sample-rate.
Limiting the sample rate to 96 kHz but with UGPM off I get the failure to record tracks as played. With UGPM on and 96 kHz I still get the white noise burst at the start of a track same as option 1.
option 2 gives zero playback issues but fails to record last tracks in play queues which triggers my OCD badly.
Option 3 works but I need to work on my “coping” bias by convincing my brain that age related hearing loss will mean I can’t hear a difference in a couple of years anyway…..
As I get older my tolerance for fiddling with computers gets less and less……..
Protect and nurture your hearing through all means… Silence is golden…
We create the hassles that hassle us… Keep It Simple…
My USB-centric Audirvāna Studio v3 Beta (2.99.20) playback system gives me much enjoyment without hassles…
Thinking about your playback scenario using the Devialet speakers as a stereo pair… If you are connecting both via Ethernet, how can Audirvāna determine which speaker is the UPnP renderer? If the Devialet App intermediary software is involved, wouldn’t this also be suspect? It seems to me that the Devialet speaker stereo configuration is the limiting factor in the context you are trying to force onto the Devialet system.
Note: I see there is a ‘Latency control’ that adjusts the audio buffer for better stereo stability.
No idea how the DOS3 software works internally to manage the 2 channels. Yes there is a latency control in the app. Increased it from the default 350ms to 1000 ms and also tried 500 ms but no changes in behaviour unfortunately.
It seems logical that the speakers communicate via a wireless protocol with each other when paired… there is no physical connection between them… Which poses the obvious question, how does Devialet define a networked set of speakers and how do they implement a stereo pair via UPnP?
I assume when you create the stereo pair the software assigns one to be the primary speaker to control processing of files. The second is assigned as the dumb secondary speaker. Only issue this system has compared to the KEF’s is that KEF uses a direct Ethernet interspeaker connection. With the Devialet’s you have the connection relying on a switch (in my case inbuilt into the wireless router) in between the two speakers.
will try with an unmanaged switch between the two instead of the router switch to see if this resolves anything. Not hopeful.. The white noise issue and or stuttering always starts at the beginning of a track so I am pretty convinced it is something to do with interpretation of gapless information in the file in AS that triggers issue but seeing as I have no idea how AS processes the gapless information in the workaround option I might be totally wrong.
So, It would seem because of the nature of a switch, there are always interrupts… I don’t see how this works reliably at all… and a router is switch, so Audirvāna will never have a single static IP address for the UPnP renderer when the two speakers are on separate IP addresses.
If it were within the switch it would occur with all files irrespective of gapless processing. It only occurs when gapless workaround in Audirvana is turned on and again only with 24/192 files not any lower resolution. Way above my brains capability to resolve.
Yes but it only ever occurs when the gapless workaround in AS is processing the gapless information. With this option set to off it never occurs. Gapless playback mode is triggering the issue.
And to illustrate the issue I have with Gapless Playback Mode turned off I get no white noise, no stuttering just glorious sound quality from the Devialets albeit without a gapless capability. However I get what is shown in the screenshot, the lack of a play count or entry in the playback history for the last track in a play queue despite all tracks being played faultlessly. For a lot of people probably not an issue…… for me it triggers my OCD or to put it in layman’s terms “it really chaffs my ass”
How do you verify that you are getting stereo playback? …Audirvāna is only transmitting to a single UPnP renderer… Maybe I am mistaken, but from your description, you are connecting each speaker independently to your router…
The only way I see to have a stereo pair configuration is via the Devialet App and it appears they communicate via a proprietary wireless protocol when paired for stereo… This would allude to one of the speaker systems being the master for both.
Is it possible that the Devialet system and Audirvāna are being ‘confused’ by having them both connected to the network via Ethernet simultaneously?
Also, this may define the sample-rate capabilities of the pair…
The startup guide advises you connect each speaker individually and recommends you connect to the router via Ethernet for greater network stability. Once both speakers are setup in the app separately you use the app to create a stereo pair, assign a left and right speaker. The app then creates the stereo pair. As regards how do you know, simple, listen to the separate guitar parts and other instruments on your favourite tracks coming from the different channels……..
Yes… I understood this…
The speakers are set-up and paired wirelessly via the Devialet App. However, Devialet does not specifically state to keep both connected to the network for stereo stability (This only makes sense to me if using Roon where multiple endpoints can be managed)...
It makes most sense to me, that one of the speakers is the master-controller of the pair and only the master-controller needs to be connected to the network… Otherwise there would be no way to synchronize the data being received by both… Audirvāna does not support a multiple endpoint UPnP transmission scheme…
Have you tried disconnecting the Ethernet feed on one of the speakers?
It would make sense to reset both, re-establish the pairing, and then connect one of the speakers with Ethernet to the network to see how this works.
Or…
Reset both… establish the network connection with one of the speakers and then pair the other for stereo (not connected to the network)…
If you think the Devialet App is controlling how the Left channel and Right channel signal data is distributed, how does the application know anything about the UPnP signal being sent to the speaker(s)? If there is some sort of feedback loop between the speaker(s) and the App, the latency would be suspect… But I doubt there is any such communication between the App and the speaker(s).
The risk of latency issues is of course greater with the configuration of the Devialet’s which is why they selected Gigabit Ethernet controllers for the Phantoms. The KEF’s which have only one dedicated Ethernet network connection (both LS50 Wireless II’s and my former LS60’s) have only 10/100 Mbps Ethernet controller. The significant improvement in latency and bandwidth of the Gbps controller negates this risk.
All well and good, however, there is no specific instruction to have both Devialet speakers connected to the network via ethernet…
In the context of Audirvāna UPnP playback, it just does not make sense to have them both connected to the network… Even when the Devialet App is the ‘server’.
And there is no specific instruction not to. Most wireless speakers limit file transmission sizes to 24/96 files when using a wireless connection between speakers (KEF downsamples internally to 24/96 with a wireless interspeaker connection, with the wired Ethernet connection this is increased to 24/192).
Further evidence of zero file transmission issues is that the Devialet play all 24/192 kHz files perfectly with wired Ethernet connections. The only time there is an issue is when the software module for Universal Gapless Playback Mode within AS is turned on. Whilst neither of us knows how Audirvana implements it’s gapless playback mode I am pretty sure the processing of this is done and the files loaded into RAM prior to any network transmission of the files taking place to the Devialets.
Let’s just agree to disagree on this one.
Update:
Just one more point about connection of a stereo pair, from the Devialet Forum…..
“For the most stable, reliable stereo performance with Devialet Phantom speakers,both speakers should be connected to your router via Ethernet. While one Ethernet cable + one Wi-Fi is possible, dual-wired connections prevent synchronization issues, audio drops, and high-bitrate streaming issues”
Also in dealing with Devialet support on this issue I have given them specific details on the network configuration with no suggestions being received to utilise a split wireless and Ethernet setup.
The question remains… How does the Devialet system discriminate between Left channel and Right channel data it receives via UPnP, in a stereo configuration where each speaker is connected to the network switch? Audirvāna has no capacity of knowing which speaker is the Left channel and which is the Right channel, as the UPnP stream is only blessed with the digital-audio packet data… How in the world can the individual IP data streams be synchronized without some form of feedback loop between the speakers? If this is the case, the latency is most likely unacceptable… Heck, the small latencies induced by USB send/recieve request interrupts can cause issues in playback performance… Something doesn’t add-up in this configuration scenario…
