Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[etc6849]

I have a decent stereo tri-amp setup now:
JRiver->RME HDSPe AES->Xilica XD4080 processor->Benchmark AHB 2 amps for mid and highs and an Emotiva XPR-2 for lows.

I want to eventually use a software option for the FIR and Bessel filters instead of the Xilica, and then use more of the AES outputs from the RME card to feed multiple Benchmark DAC3 DX's. I want to do this for all surround channels and also have everything below 100Hz go to my dual subs (and need the flexibility for me to adjust timing, filter type and slope as needed).

Needless to say, tri-amping all my surrounds is an expensive endeavor and I need to be sure that:
1. This will work for up to 5.1, which would use all 16 AES output channels (5x3 + LFE/Sub) of the RME card
2. That I will have the versatility to combine the two 8 channel RME WDM devices using some software option (RME lets me pick between 8 stereo or two multichannel WDM devices in how I use the 16 channels)
3. That such a solution exists for Windows 10
4. That JRiver rerouting would work, so presumably I could still use Netflix or PowerDVD for 4k UHD playback?
5. That this would not mess room correction up as I use the Dirac VST plugin now from within JRiver?
6. I would assume that processing would look like:
JRiver 8 channels out -> some audio processing software -> output to the 16 AES outputs?

My other idea was to use more than one Xilica XD4080, and I ran with this for a while where the second one just handled delays and crossover to my subs for the surround channels.  It worked well, but I want the best possible performance.  I like the appeal of a software option as this could yield better performance, and let me pick/swap my DACs out.

[mattkhan]

>> RME lets me pick between 8 stereo or two multichannel WDM devices in how I use the 16 channels)
if each card presents as a separate device then that is a problem for jriver as it only supports a single output device per zone. You would therefore need to create a virtual device to aggregate the underlying devices (e.g. using asio4all). Other audio devices, including other RME devices, expose the additional hardware as part of the same master device so don't have this problem.

>> JRiver 8 channels out -> some audio processing software -> output to the 16 AES outputs?
assuming you get the above to work then you would simply use one of the higher channel count output formats in jriver. This will result in a 7.1 target though so you will need to decide how to deal with this (e.g. don't use jrss, manually process 7.1 content). Dirac doesn't sound ideal for this job though as it doesn't deal with crossovers as part of correction, audiolense or acourate seem like better options. If you do want to stick with dirac then you'd need to apply room correction before the crossovers. How do you intend to implement the crossovers? via convolution or PEQ?

>> 4. That JRiver rerouting would work, so presumably I could still use Netflix or PowerDVD for 4k UHD playback?
by rerouting, do you mean the jriver wdm driver? if so it works fine. You may find latency is an issue for video playback though.

[etc6849]

Hi Matt.  Thanks for your help!

For WDM devices, there would be two 8 channel devices, but for ASIO it seems REW can see 16 output channels   Now, I don't know how to verify JRiver can see the same, I would assume it can since it uses the same RME ASIO driver.  The WDM limitation shouldn't matter, assuming JRiver's virtual sound device can show up as a WDM device and allow direct sound, but use ASIO in the background?  Attached is what REW is seeing (without using ASIO4ALL).

I would like to use FIR filters like I have been with the Xilica as they sound very good to my ears (versus higher order linkwitz-riley, etc...).  Anything other than FIR filters seems like a compromise (and why not use them as my PC is very fast).  With the Xilica, I was able to setup my system with only 180 degrees +/- phase shift from ~20Hz to 20kHz.  I know most say you can't hear phase coherence, but I'd swear it sounds more life like and the bass blends pefectly into the image now.  For crossing to the subs, I like to use a low order bessel and swap my sub polarity as this works best keep phase very flat.  However, maybe a PC will not have the tap limitations the Xilica had, so I can use FIR for subs too?

I looked at Acourate's site, but for the life of me couldn't tell if it did FIR XO's and would let me do an actively quad amped setup (tri-amp + subs).  Same with Audiolense's site.  Assuming I want a mix of FIR XO's and low order filters for lower frequencies, plus room correction, which would be the best option to try?

Exactly right.  Dirac doesn't do crossovers and this is why I am looking at other options.  I have had issues with ASIO4ALL before, but my PC RME's ASIO driver (aka hammerfall) works perfect and sounds much better than windows mixer.  It would appear from the attached that all 16 channels show up as a single device when using the RME ASIO hammerfall (not so when using WASAPI or direct sound though).

>> RME lets me pick between 8 stereo or two multichannel WDM devices in how I use the 16 channels)
if each card presents as a separate device then that is a problem for jriver as it only supports a single output device per zone. You would therefore need to create a virtual device to aggregate the underlying devices (e.g. using asio4all). Other audio devices, including other RME devices, expose the additional hardware as part of the same master device so don't have this problem.

>> JRiver 8 channels out -> some audio processing software -> output to the 16 AES outputs?
assuming you get the above to work then you would simply use one of the higher channel count output formats in jriver. This will result in a 7.1 target though so you will need to decide how to deal with this (e.g. don't use jrss, manually process 7.1 content). Dirac doesn't sound ideal for this job though as it doesn't deal with crossovers as part of correction, audiolense or acourate seem like better options. If you do want to stick with dirac then you'd need to apply room correction before the crossovers. How do you intend to implement the crossovers? via convolution or PEQ?

>> 4. That JRiver rerouting would work, so presumably I could still use Netflix or PowerDVD for 4k UHD playback?
by rerouting, do you mean the jriver wdm driver? if so it works fine. You may find latency is an issue for video playback though.

[mattkhan]

For WDM devices, there would be two 8 channel devices, but for ASIO it seems REW can see 16 output channels   Now, I don't know how to verify JRiver can see the same, I would assume it can since it uses the same RME ASIO driver.  The WDM limitation shouldn't matter, assuming JRiver's virtual sound device can show up as a WDM device and allow direct sound, but use ASIO in the background?  Attached is what REW is seeing (without using ASIO4ALL).

that's fine then, just use the ASIO device as your output in jriver and set output format to 16 channels

I would like to use FIR filters like I have been with the Xilica as they sound very good to my ears (versus higher order linkwitz-riley, etc...).  Anything other than FIR filters seems like a compromise (and why not use them as my PC is very fast).  With the Xilica, I was able to setup my system with only 180 degrees +/- phase shift from ~20Hz to 20kHz.  I know most say you can't hear phase coherence, but I'd swear it sounds more life like and the bass blends pefectly into the image now.  For crossing to the subs, I like to use a low order bessel and swap my sub polarity as this works best keep phase very flat.  However, maybe a PC will not have the tap limitations the Xilica had, so I can use FIR for subs too?

you can use long FIR filters on a PC, hence can use them at sub frequencies, however the latency is substantial (e.g. ~660ms for a 65536 tap filter) so is not suitable for live video playback. If jriver is playing the video then this isn't a problem but it won't work with UHD via WDM. You *might* be able to use 16k filter length (so ~160ms latency) or 8k (~80ms) if your display has substantial latency (e.g. a projector), something to experiment with. Alternatively you can experiment with non linear phase filters.

I looked at Acourate's site, but for the life of me couldn't tell if it did FIR XO's and would let me do an actively quad amped setup (tri-amp + subs).  Same with Audiolense's site.  Assuming I want a mix of FIR XO's and low order filters for lower frequencies, plus room correction, which would be the best option to try?

certainly acourate can and AFAIK audiolense can. I think the generally accepted view (NB: I have no personal experience of audiolense so just going on comments I've read) is they provide similar SQ but audiolense is more user friendly, particularly for multichannel, while acourate is more flexible. I use acourate btw, there are other posters here who use audiolense (e.g. mojave).

[etc6849]

@mattkhan

Thanks Matt!

No live TV, so delay is fine although my LG OLED delay is minimal.  However, you are right about Netflix and Amazon and any external program beside JRiver handling video playback being affected...  Didn't think about that until I read your post, doh!

I will have to consider ditching Netflix, UHD Bluray and Amazon Prime video.  So far, I've been using two Xilica XD4080 units and they are very transparent, but only support up to 96kHz and a total of 3600 FIR taps.  Since the total FIR taps are limited, I'm only using FIR filters at 500Hz and 3.8kHz.  For the subs to woofers, I use low order Bessel filter and swap sub polarity (and also plug the ports on my towers).

Have you tried using linear filters and then all pass filters to correct phase, or is the auto room correction in Acourate good enough to where it flattens phase very well for the final output (e.g. corrects any phase shift that was added by classic linear filters like BW, LR, etc...)?  Also, how does Acourate handle multiple sampling frequencies (fs)?  I am assuming it generates filters for each fs like Dirac?

In Acourate, for the FIR filters is it just a matter of inputting how many taps to use and the cutoff frequency?

PS: I have my Benchmark amps on order   Waiting for more funds, but is the Benchmark DAC3 DX really the best possible option for the DACs? 

8 of these DACs are going to be very pricey, but I am wanting the best possible fidelity.  I just need a DAC with a very low noise floor that takes AES as an input and outputs fully balanced signal via XLR.  Maybe the DACs in the Xilica XD4080's are good enough to where I won't see an improvement in using a DAC3 DX for each channel pair?

Certainly gain matching of the Benchmark amps I have ordered will be perfect if I use benchmark DACs, but I am getting into sport car pricing.  My end goal is to have the best system at any price though.

[mattkhan]

Have you tried using linear filters and then all pass filters to correct phase, or is the auto room correction in Acourate good enough to where it flattens phase very well for the final output (e.g. corrects any phase shift that was added by classic linear filters like BW, LR, etc...)?  Also, how does Acourate handle multiple sampling frequencies (fs)?  I am assuming it generates filters for each fs like Dirac?

the choice is yours acourate has a set of macros that perform a semi automated (room) correction, this performs separate magnitude and phase correction steps based on your defined target curve and where each correction is performed according to a number of parameters which you control. These corrections are combined with your crossover filters and you can embed other things in these (e.g. a driver linearisation filter, inter driver delays). The crossovers themselves are also generated in acourate and can be minimum or linear phase, alternatively you can import one created elsewhere.  The final filters can be generated in your choice of format and sample rate(s).

Aside from this, you can use the toolbox features to construct all sorts of things. To give an example, the filter I generate for my near field subs embeds a VBA (virtual bass array) filter (basically a delayed, inverted copy of itself to soften a room mode without impact the direct sound), a steep low pass filter at 47Hz (to avoid the NF sub contaminating the audible range too much) and a custom phase correction filter to "remap" it's phase response onto the correct phase response of my main subs (to avoid any unwanted interaction effects as the NF rolls off).

I then simply load those filters into jriver via a cfg file and away we go. It's not a user friendly app by any stretch so the learning curve is quite steep, great if you like to tinker though. https://www.amazon.com/dp/B01FURPS40 gives a good account of how to use it.

[etc6849]

Thanks!  I will purchase this book and study it.

This virtual sub idea is very intriguing and clever!  I would think it would work very well.  Do you typically do room correction filters after after averaging many points like Dirac does (for the crossovers however, I'd plan to do a single point at the MLP)?

If you know someone as well versed as you in the Columbia, SC or Atlanta, GA area, maybe we could trade services as I know a lot about home automation and I'm a decent HA programmer.

I then simply load those filters into jriver via a cfg file and away we go. It's not a user friendly app by any stretch so the learning curve is quite steep, great if you like to tinker though. https://www.amazon.com/dp/B01FURPS40 gives a good account of how to use it.

[mattkhan]

This virtual sub idea is very intriguing and clever!  I would think it would work very well.  Do you typically do room correction filters after after averaging many points like Dirac does (for the crossovers however, I'd plan to do a single point at the MLP)?

acourate is a single point measuring system when using its macros, though you can design your own filters as well using whatever process you like (e.g. moving mic method or by designing sub filters in MSO). IME the single point method does work very well.