Years ago on another forum, Tir Na Hifi, a user named "jesuscheung" began posting a lot of Windows-based tweaks and optimizations (then mostly for Win 7 and Server 2012 and the like), where myself and others would contribute. "JC," as we came to know him, was influenced by a user in China, "pkshan," who had a Chinese language-based blog and website where his ethos was tweaking Win 7 and foobar, but based purely on SQ, not on minimizing everything. So, in other words, he would make one (seemingly meaningless or unrelated to SQ) change, like change a transparency setting for each window, or change the size of icons on the desktop, and after each change he would listen to see if he felt like this sounded better or worse. At one point, pkshan created a whole ISO for Windows 7 with many such changes, albeit only available in Chinese. "JC" applied this ethos to his own Windows tweaks, where he would try something, then listen, and regardless of whether it meant reducing latencies, or CPU activity, or memory usage, or whatever the case might be, he would reverse the change/tweak if he felt like it sounded worse. You can find the thread here, many years dormant -- still has some interesting info!:
https://tirnahifi.or...419cea6c0ae28a1
Anyway, I bring this up because what kuni is describing with some of the services and settings sounding "better" or "worse" reflects a similar approach. This is a totally legitimate way of approaching tweaking/altering/"breaking" Windows to achieve better sound quality, and over the years I have seen some combination of one of three scenarios for doing that:
1) minimizing Windows footprint above all else, whether that means reducing latencies, CPU activity, RAM activity, or simply reducing Windows size. The theory driving this is that the smaller/more broken/the less footprint you achieve in Windows, the less "noise" is generated by the OS and, concurrently, in the music server hardware, and/or latencies are improved, all resulting in better SQ.
The first example of this I ever encountered was on "Computer Audio Asylum" where the cMP ("cics memory player") project began in the 2000's, which eventually culminated in users reducing Windows XP to a ridiculously small size, something like 20 to 30 MB. There were various registry changes involved as well as other of the kinds of steps we do even now (turning off services), but there was also a very heavy focus on manually deleting system files in places like the c:\Windows directory, which is a very slow and methodical approach with a lot of headaches, but they were extremely successful. More successful than many modern attempts in terms of sheer size, at least, although some of that is reflected in the already smaller size of Windows XP before modification vs. modern Windows OS'es, even PE.
2) tweaking as I just described, meaning making a change to the OS in some way, listening, and keeping or rejecting the change based on SQ alone
3) some combination of the above two, which many resort to -- this is more what kuni has resorted to
Anyway, personally, the approach of listening to each tweak or change just drives me too nuts, lol, and I have long ago committed to just reducing Windows to as small of a footprint as possible, despite what may happen to SQ for each change. I also have become much more dependent on solutions like WinXPE's or using something like WinReducer/NTLite on a "normal" Windows ISO, meaning that much of the reduction in size and bloat occurs before the ISO is even created and installed/booted. For a modern Windows OS like Win11, this is much preferable to manually deleting portions of the OS post-install, as it also reduces traces of the "packages" removed from registry and elsewhere (there's some evidence that registry size also effects SQ, with bigger being worse).
Taken together, all of these changes toward reducing size and footprint have usually, at least eventually, resulted in "better" SQ for me, and has helped me keep my sanity versus listening to every change. I also would say that reversing changes that "should" sound better is fine, but it is probably best to test in each of our systems, as the synergy of various OS changes and all of our different components and transducers etc. will give us all different results.
Regardless, Windows PE has allowed me to get Windows down to the least amount of CPU activity (as measured in handles/threads), memory usage, processes, and services of any version of Windows I've ever tweaked/modified. It also is much easier to do things that are either impossible in other versions of Windows (terminating lsass.exe, for instance) or require a ton of work to do so.
Kuni, I am unsure why WinXShell is not working for you in Win11PE. Do you mean it just doesn't boot at all, or that programs aren't running, as you say? Which audio players do you mean?
