Yes, that’s the case. I know people have got it running. But since it goes under category “advanced nerdy stuff” I don’t make any promises about it, and I don’t provide any help setting such things up either. But I don’t have reason to block it either (only pre-Vista Windows versions are blocked).
As such Windows Server is not optimized or intended for multimedia use, rather it is optimized to perform well as file/database server on a network. The two use cases have largely opposite requirements in many respects, for example one can optimize OS for I/O throughput or latency. Servers are optimized for I/O throughput at the cost of latency while desktop OS are optimized for multimedia and thus for latency (interactivity) at the cost of I/O throughput. It may be possible to tweak Windows Server to perform differently than Microsoft originally designed, but that goes beyond my scope. There are other aspects as well, such as desktop OS are usually designed to have more eye candy which costs CPU/GPU time while server OS are not designed to be pretty saving CPU/GPU time for crucial tasks. HQPlayer itself tries to balance it’s amount of eye candy more on the server paradigm in the desktop GUI mode, and on the desktop paradigm in the fullscreen/touch mode (sounds confusing, doesn’t it?).
So with AO and some other extra work it may be possible to transform WS2012R2 to work nicely for HQPlayer, but I hope people understand that such complex setups are not feasible to be supported from my perspective.
Yeah, I have no intention to pay someone not to be supportive.
Moving on, one option is to use Acourate Convolver to receive the stream from Roon, convolve, and send the multichannel streams to each channel in the DAC. Acourate Convolver would be installed in the audio PC (where I try to minimize the number of processes being run). What would be the best Roon configuration for this?
Have Roon Core in another computer and have it stream to the audio PC? I’m guessing this would minimize the workload on the audio PC.
If you ran Roon Bridge in the audio PC then that would get the signal from Roon to the audio PC via RAAT, but I’m not sure how you might then send a signal from Roon Bridge to Acourate Convolver in the audio PC. If Accourate Convolver can take an input signal from the audio device on the audio PC then that might do it, but I haven’t heard of anyone doing anything similar. It may be that others can assist.
I sense this might be able to solve part of the jigsaw puzzle but am yet to try it (I was looking at local loopback to shoehorn in Dirac, but didn’t have time to experiment.)
Have you tried Audiolinux? Somebody that used to run WS2012 plus Optimizer said he switched to Audiolinux, that comes optimized with Roon and HQP, and he says it sounded better in his system. Maybe that’s an alternative path for me, but a major change in my system.
Thanks, I’ll look into it. I have partitioned the drive on the BRIX to run a Linux, but haven’t done so because I know if I change over I’ll stuff something up and be without music for a few days. I also find Remote Desktop really good to control HQP from an iPad, not sure if there is a Linux to iPad equivalent.
Do you think it’s a bad idea to run Roon Core on the WS2012 audio PC? It would be the simplest to implement, but I wonder if the intensity of Roon processing will affect sound.
Do you know if Roon Core will run on WS2012 in core mode?
No problems about the questions, but I will pull the OT posts into a new thread later, so that people who aren’t necessarily interested in this part of the discussion aren’t bothered with it. Just want to check if there is an appropriate existing thread to take it to or start a new one.
I think the “thin client” model for the final connection (Ethernet renderer ?) to the DAC is a good one. I use the mR myself for that reason. I believe that keeping the Ethernet renderer as thin as possible with the best clean power (I am a happy user of the Up Tone UtraCap LPS-1) means that we can be much more agnostic about the server PC compared to a single PC solution. Chris Connaker, in the CA comments below his review of the SOSE, noted that he couldn’t hear any difference when using treated/untreated servers as input to the SOSE as an Ethernet renderer. I would expect no less from the mR.
Having said that I think room EQ (and some attention to room treatments, placement of furnishings, elimination/dimnution of first reflections etc.) is probably the best bang for buck improvement that we can make in the hobby. So if putting the Core on your audio PC meant you could implement room EQ then I would suggest going for it and seeing how it sounds. My guess is that the improvements from room EQ would be greater than any processing effects in the audio PC, particularly if your audio PC is otherwise isolated from power noise etc.
Thank for the answers. Helpful indeed. Been doing further reading and steering towards trying HQP. I started a new thread asking system architecture on the Audio Products forum. Hope it’s the right one. Please chime in!
While it’s not beautiful, I have found a workaround that enables me to use JRiver’s convolution engine with a TIDAL stream. I thought you might be interested since it may work for you as well. Here’s a link to some instructions that I typed up last night: HOWTO: Streaming TIDAL via JRiver Media Center.
Again, this is not a particularly beautiful solution (for example, gapless playback does not work), but it will tide me over until the good folks at Roon Software get around to adding JRiver-like convolution to their product. Cheers!
Thanks for posting this, and congrats on finding a workaround! I saw your post on the Acourate forum this morning and looked into it. Well done!!
I’m weighting trying Tidal/Roon/HQPlayer. I discovered my DAC is likely not to work with a microRendu as a multichannel NAA with HQP so that is yet another hurdle. Might try your solution.
Hi David
Thanks for your solution. It worked out just fine.Have You also tested with Roon as remote? When I tried that JRiver stopped and told that it couldn´t process 32-bit files… Somehow it doubled the bitrate.
I use Primare DAC30 with their ASIO driver.
Anyone have a comment? (ps it´s 4 am in the night here in Sweden so my brain isn´t at rocket science level for the moment ) ZZzz
Thanks for giving this a try and confirming that it worked for you. Glad to hear it!
I did try enabling the ASIO virtual driver in JRiver (similar to the WDM driver). This presents a virtual ASIO sound card to roon. I then enabled that device in roon, but JRiver complained about the number of channels…if I recall correctly, JRiver thought that it should be receiving 16 channels of music while roon was only sending two. I might be remembering that wrong, but in general, the mismatch prevented JRiver from playing. When I had JRiver downmix the 16 channels it was seeing to stereo, the CPU in my little PC was not powerful enough, so I would get a second or two of sound and then things would lock up.
I also tried configuring roon to play through a demo of Acourate Convolver. There were no problems with channel mismatch, and it worked better, but I still had trouble with dropouts every few seconds because my PC was not fast enough.
The DLNA solution is the most reliable for me so far. I have enabled the “Play files from memory instead of disk” and “Load files to memory at the start of playback” options in JRiver to ensure that things reliably. The only downside with these options is that it can take 10-20 seconds for playback to start because the PC is reading entire file from my tablet over wi-fi. There are also usually several seconds of silence between sequential tracks on an album. Not brilliant, but far better than having dropouts and skips in the middle of each track!
Again…really hoping that the product managers here at Roon Software will get JRiver-style convolution added to the roon product roadmap, but I’m up and running with JRiver and BubbleUPnP for now.
I moved this thread to the “feature requests” section. Some work is underway for 1.3.
The parametric equalizer is “alpha quality” already–barring some significant unforeseen event, it will be part of 1.3.
A convolution engine is in progress. We have a prototype, but it needs some performance work and a few glitches ironed out. I’m not 100% sure that it will be fully baked by 1.3, but we are going to try. If it doesn’t make it in time, rest assured that it is in our plans, work is in progress, and it will come as soon as we can manage it.
It would be very helpful if anyone using convolution currently could respond with the answers to a few questions:
What software do you use to create filters?
What software do you use to apply filters?
What do you use the filters for (e.g. room correction, headphones, digital xo, …)?
How large are your filters (how many taps, how many paths)?
What CPU are you running them on?
I may send PM’s and ask some of you to send your filters/configurations so that we can build a representative sample of “real world” configurations for development/testing.
I am using REW’s generic equalizer for three parametric filters. Export one mono impulse wav file at 24bit/192kHz, normalized. Wav file used for left and right channel convolution in HQ player.
I use convolution for room correction.
Computer for REW and HQP is a headless Mac mini 2012 i7 with 16 Gb, running Roon core and end point as well.
I use Acourate for measuring my room and creating the filters. I use either Acourate Convolver or JRiver to apply the filters. With Roon, I use Acourate Convolver, as it presents a virtual ASIO driver that Roon can stream to from the Roon Server on my Win12 CAPS Server (i7 w/ 16GB RAM). I use this for room correction only.
I’d be happy to be involved with beta testing, as I already have Roon running with AC, so I can do an A:B comparison.
I use Audiolense for measurement and filter creation. Most of the time I use LMS with the BrutefirDrc plugin. As Audiolense defaults to a standard wav file for stereo. I use sox to split it into two raw PCM files L and R for brutefir. The Audiolense stereo wav can be used directly in JRiver, much less hassle than brutefir.
I’ve used 132K taps and can convolve 24/192 (barely) on the (linux) NAS where my music is stored. This is powered by a Intel Atom D525. Running the same on a Intel i5 NUC (also linux) is much better allthough I believe that if the music files where local on the NUC this would be even better. JRiver running on W2012 R2 on the NUC (as UPnP server) works very nicely too.
I use stereo full range convolution for room correction, but also a little parametric EQ in the bass range.
) ZZzz