Okay I got check with my Diablo and borrowed a pro iDSD sig to do a tonality, density, macros check on the new Soekris firmware and they practically match. I always considered the previous Soekris firmware to be mid-centric and it was still giving a lot of bass to the Stealth so I didn't mind it. What I do miss a lot was the kick drum realism of the previous firmware. There was a lift in the upper bass so it gave just the right amount of emphasis, air and reverb that made me impressed with the kick drum presentation. There's a bit less sustain now and it decays even blacker now so it is what it is. Other than that, this is the best setup combo I've heard.
So much so that I learned a lot on this combo. Few years ago I was trying to optimize a listening room with Dirac 2.0. The end result, it gave a very headphone listening kind of sound. I just went with the flat FR preset. With the Stealth and Phonitor One, I hit gold with using crossfeed. It just gave that extra help in realism. I wasn't that convinced with the older soekris firmware + xfeed, but it was like I passed through a portal with the new firmware plus crossfeed combo. So it made me think what's going on. I even prefer a lot of qualities on it compared to my Dirac calibration on the listening room.
So here's my theory, my Dirac calibration was to make loudspeakers in a room have a flat response. So it's speakers + room then making it a flat response. This Stealth setup though, it's a Harman curve response plus very low distortion and group delay then adding crossfeed on top. It felt more like a listening room than my Dirac calibration.
And if you're following, this starts to make sense. The Dirac calibration cancelled out , or didn't lend to a natural room response that would make us feel it's a band in the room. I wasn't keen on the proper amount of mid bass that time so the flat response felt more open and more faithful. I tried a slight downward curve and some Harman preset but it didn't interact with the room well so it felt like the worse option at the time.
No crossfeed is a tighter presentation especially on electo synth bass. Great if you love that punch, purity and density of it. With crossfeed, there is a bit of sweetness and air around the attack. Less fatiguing. Vocals are sweeter and more convincing too. Without crossfeed it has the accuracy, purity, density of iems. With crossfeed there's a contour in the vocals that's pleasant and convincing.
What's very new also is how vocal peaks are presented now with the combo with or without crossfeed. Typically when a singer starts to go strong on a word, the presentation moves forward, becomes sharper, hard and louder. With the current combo, the peaks move a bit back instead and the detail lines up in a row and is very drawn out, relaxed and coherent. The same goes for very busy strings/orchestra passages. Nothing clumps up. There is enough time and space to keep everything in a row and given their chance to speak.
Oh with the pro iDSD sig, unity gain was not loud enough for the Stealth to my surprise. +9dB gain killed the blacks, bass and dynamics so much that i wouldn't recommend the pairing. 4.4 unity gain may be sufficient but I loaned my 4.4 out so I can't verify atm.
Here's a link to my LCD 5 impressions and how it compares with the Stealth:
https://www.head-fi.org/threads/audeze-lcd-5-review-measurements-interview.959808/post-16790071