i feel 2.44x is good, soundstage have stable, bass vibration correct, and other things..
optimized code, not so much better, but sound texture more correct.
uploaded v2.45 160, v2.45 256, v2.45 352, v2.45 448 with different buffer sizes, maybe help you better adjust system clockrate by variuos options.
in my case v2.45 160 is best, good vibration in good soundstage. v2.45 256 is second. i use default Pro Audio setting 10000. try and enjoy!
for my soundcard, 160 sometimes is better. but treble tune overdone a little same as XA. long time listen ears can soar.
please also make 356 and 444.
my soundcard gives best vocal with 356 (352+4). maybe vocal detail can beat XA. (apart from vocal vibration and -0.5% emotion, XA is pretty perfect in vocal)
for my setup, vocal is coming alive with 352! more micro-detail than 256. warmer than 160.
Last edited by jesuscheung on Mon Dec 09, 2013 12:59 pm, edited 2 times in total.
448 3D soundstage. -2% details. energy and power. fatigue...
352 actually is the best after mod. bad without mod. max detail(can hear saliva. cannot with 256). simple soundstage.
256
160 just treble tune not good. overall good.
maybe 356 best for me. only maybe
Last edited by jesuscheung on Mon Dec 09, 2013 2:30 pm, edited 1 time in total.
jesuscheung wrote:lekt, do know which dll i need to mod for wasapi? in system32, i see
audiosrv.dll
audioses.dll
aduiokse.dll
audioeng.dll
audiodg.exe
i don't see audioendpointbuilder.dll.
which of these files are wasapi?
need some learning them, most important that what functions render loop calls.
WaitForSingleObject(...); KERNEL32.dll
pAudioRenderClient->GetBuffer(...); ??
pAudioRenderClient->ReleaseBuffer(...); ??
where are GetBuffer and ReleaseBuffer locate? GetBuffer return 8 values (there's 8 checking for error/non-error), ReleaseBuffer return 6 values. if wasapi does'nt check for these 14 values then SQ will much perfect, latency will reduced so much. re-write wasapi code if you MS developer, hehe... will do learn more.
sbgk wrote:can't find enough info to get started. the kssample doesn't work, so would need to look at portaudio code to understand what's required.
Thing is I've never been that impressed with the sound of KS when I've tried it, sounded a bit soft. It was interesting reading about the different modes KS can use eg memory mapped between source and destination so there is no copying required, think this could be used in soundcards. USB devices require the data to be copied, so that puts a ceiling on how good usb can sound, perhaps. Am going to demo a NAD M50, so will be interesting to hear what a high end streamer sounds like, might be good enough for me to give up on MQn.
it means in KS case, KS does'nt provide data buffer, client creates it and set pointer for each data package in loop. excelent! sbgk. wasapi exclusive hold data buffer, can not touch it, only copy data into it. need also get its pointer for each iterator in loop. bad idea, think that.
don't give up with KS. i'm waiting new wasapi version, hope wasapi will let do more with data buffer.