Hi All, Iām long overdue an update on the active side of things. Despite only properly working 3 days a week now, my free time has been limited. When you add in the fact that the learning curve with Acourate looks like a Heaviside step function, progress has been slow!
However, as of yesterday, I have a working filter set. Itās not perfect and needs some refinement, but itās opened a window, more on that later.
My original intention was to set things up for dual TV and music use, however something dawned on me the other week when I was tinkering - a 131072 tap FIR filter at 48000Hz adds a latency of 1.37 seconds 
Well beyond the lipsync adjustment range of domestic A/V gearā¦ Yes there are technologies and hardware that can delay video by that amount, but theyāre beyond the realms of sensibility for me.
So I bought SWMBO a Sonos Arc Ultra soundbar for the TV. Sheās delighted, loves the sound and is happy that only one remote is needed to control TV and sound. I can now revert back to USB input from Roon just as soon as Iāve sussed out VB Audio Matrix and can connect the processing PC, Roon, Hang Loose Convolver and the Motu Ultralite Mk5 together and get them all to play nicely! 
Hereās a rough diagram of the setup:
Not shown in the diagram is a Bobwire DAT1 which takes an optical output from the Motu and via a splitter, turns on all of the amps as soon as a music signal is detected.
The RPi running Ropieee is still connected via TOSLINK. Eventually, the PC will have Roon running on it and the RPi can be repurposed as a display/remote or another endpoint elsewhere in the house.
The PC runs Mitch Barnettās Hang Loose Convolver and handles the ASIO drivers and CueMix5 application for the Motu Ultralite. HLC is a really cool application - it allows seamless on the fly switching between up to 6 different filter banks. So you can blindly (if you get someone else to switch them over for you) compare different filter sets to see what you prefer. In non-multichannel filter sets, thereās a bypass button to compare, say a room correction filter on and off. I could have gone with a multi-channel DAC like the Okto dac8 PRO, but originally went for the Motu under the advice of a friend because I originally wanted to have TV and hi-fi audio inputs. Even now that itāll only be the hi-fi, Iām still glad I went with the Motu. I can swap the USB cable to my laptop, plug in my Earthworks M23R mic and Iām all set to run whatever measurement I choose on any/all of the 6 connected output channels with Acourate without having to mess around with loads of cabling. The Motu has 10 line level (+21dBu / 8.7 Vrms) TRS balanced outputs so thereās plenty gain to play with in correction filters and enough outputs for subwoofers for later onā¦ It also has almost impeccable measurements for itās ADCs/DACs and mic preamps.
For the DSP side of things Iām using Dr. Ulrich BrĆ¼ggemannās Acourate. Itās pretty unique and more than a little complicated. So much so that Mitch Barnett wrote a book on how to use it.
Most of the other DSP utilities for active crossovers & room correction range from the equivalent of a takeaway dinner or a ready meal to a box of ingredients with a recipe delivered to your door. Acourate gives you a larder of ingredients and a kitchen in which to prepare and cook them. Not only that, it lets you taste the meal at the end and then go back and take out some spice, add a little salt, cook it for longer (or shorter), chop the ingredients more finely, and taste again. Once youāve done the basics, itās fairly quick to go back and try something different.
Firstly, you start with a crossover design and you can choose from Butterworth, Linkwitz-Riley, Bessel, Neville Thiele, Horbach-Keele, and Dr. BrĆ¼ggemannās proprietary āUBā filters. The UB jPol11 filters are really interesting in that they have very steep slopes and minimal pre-ringing.
Next step is linearising the driversā amplitude and phase response by measuring each driver individually over the frequency range needed
And then you convolve the crossover curves with the individual driver responses
Next up is time alignment. The distance from each driver in a speaker to the ear is different, which results in the soundwaves from each driver arriving earlier or later relative to each other. The differences can be very small, but theyāre not insignificant. Poor time alignment affects clarity, can smear the sound and distort the timbre. This is done by looking at the peak of the impulse of the bass or midrange driver relative to the tweeter and rotating the crossover by the number of samples difference. My midranges were 1 sample ahead of the tweeter (0.02 ms or 0.7 mm distance) and the bass drivers were 5 samples behind (0.1 ms or 3.5 mm distance). It doesnāt sound much but humans use interaural time differences as low as 10 Āµs to locate the direction of sounds, so itās not unreasonable to assume that very minor time delays would affect the positioning of sounds in the soundstage and in reality, they very much do.
After time alignment, itās a case of generating an initial filter .wav, making logsweeps of left and right, then proceeding with the room correction. You can design a target curve or import one which is then inverted into the initial response, then itās a matter of generating the filters, making a test convolution, looking at step response, group delay, pre-ringing and iterating backwards and forwards between filter generation and test convolution until youāre happy. Next is inter-channel phase alignment, followed by going back to filter generation with your tweaked recipe and then creating the filters in your desired format and sample rates for your chosen convolver:
You can then generate a verification .wav from the final correction filters to measure your L & R & mono responses. The mono response might show some anomalies, which you can then tackle in a variety of ways. I had a peculiar dip in the lower bass which I was able to correct (with some help) via a manual reverse All-Pass filter.
I still have quite a bit of tweaking to go, but what Iām hearing with the linear phase, fully active, room-corrected crossover compared to the original passive crossover corrected with REW and rephase is literally night and day.
The soundstage has grown in all 3 dimensions, clarity is vastly improved, positioning of instruments and vocals within the soundstage is much more defined and the decay of a guitar string for example is utterly lifelike. Closing my eyes on one track, I would swear there was an acoustic guitar being played in my living room about five feet in front of me and about a third of the way between the right and left speaker. Acourate has given me a new window into music listening that just doesnāt exist with a passively filtered speaker.
Iāll share some more measurements when Iāve done the next round of tweaking.
