Tír Na HiFi

i've tried some days with different compiler settings, not sure in result. compiler also is difficult thing for sound adjustment.

uploaded:
lekt.exe1644 v2.87.2 1024
lekt.exe1644 v1.17 1024 (in folder v1 (1 FILE))

JC, could you test these 2 versions, have pops in your case or not. in my case pops..., much pops.
v2.87.2 use 1.5MB RAM, v1.17 new method reading file.
try them.

lekt wrote:i've tried some days with different compiler settings, not sure in result. compiler also is difficult thing for sound adjustment.

uploaded:
lekt.exe1644 v2.87.2 1024
lekt.exe1644 v1.17 1024 (in folder v1 (1 FILE))

JC, could you test these 2 versions, have pops in your case or not. in my case pops..., much pops.
v2.87.2 use 1.5MB RAM, v1.17 new method reading file.
try them.

no pop

2.87.2 much improves on micro detail. clean.
2.87.2 > 2.87.
2.87 > 2.86.
stage seems stable.

1.17 mellower/warmer.

relisten s8. think it has cleanest very refined micro-vibrations.

s8, s4 are best of 2 files model. s4 have very rich sound, high SQ, will back to it, 2 files can give more potential coz isolate process better than isolate thread, lesser negative affect to sound.
v2.87.2 have good sound texture, details.

JC, i have transfered all things to C, and compiled to C-native. bulded 2 versions with more simple code, running with 2 threads. ah, C syntax take many my time, few days working with debug, now running, hehe...

uploaded:
v2.88 1024
v2.88.1 1024

now will make more complex code, can make sound better with C, only syntax, bugs... assembler later, need more learning.

lekt wrote:s8, s4 are best of 2 files model. s4 have very rich sound, high SQ, will back to it, 2 files can give more potential coz isolate process better than isolate thread, lesser negative affect to sound.
v2.87.2 have good sound texture, details.

JC, i have transfered all things to C, and compiled to C-native. bulded 2 versions with more simple code, running with 2 threads. ah, C syntax take many my time, few days working with debug, now running, hehe...

uploaded:
v2.88 1024
v2.88.1 1024

now will make more complex code, can make sound better with C, only syntax, bugs... assembler later, need more learning.

very nice! much room for improvement

2.88.1 feels better. treble better. cleaner micro. better emotion.
2.88 little dirty at micro, i think.
stage seems stable.

for 2.87.2, 2.88, 2.88.1, i think the width is stretched left and right. after a while, feels unnatural

jesuscheung wrote:for 2.87.2, 2.88, 2.88.1, i think the width is stretched left and right. after a while, feels unnatural

and i feel a litte earache although not intense, not strain. maybe sharpness, balance?

uploaded:
v2.88.2 1024
v2.88.3 1024

prefer v2.88.3, small file size, used new method, single thread. but pops in case of onboard sound card.
try them.

lekt wrote:

jesuscheung wrote:for 2.87.2, 2.88, 2.88.1, i think the width is stretched left and right. after a while, feels unnatural

and i feel a litte earache although not intense, not strain. maybe sharpness, balance?

uploaded:
v2.88.2 1024
v2.88.3 1024

prefer v2.88.3, small file size, used new method, single thread. but pops in case of onboard sound card.
try them.

micro-vibration:
-when sound volume is weak
2.88>2.88.2>2.88.3
-when sound volume is high
2.88.3 seems best

(i misjudged 2.88.1. it strains the sound little bit, so it has false-clean.)

when details of weak volume is needed, 2.88.3 gets unstable in stage. this is reflected in rock when lots of drums happening.
normally, 2.88.3 > 2.88.2

i see 2.79.2, 2.4.3, s4, s8 have clean micro-vibration, soundstage is automatically stable.
2.86 is odd one. little-bit dirty, but stage remains stable.

jesuscheung wrote:i see 2.79.2, 2.4.3, s4, s8 have clean micro-vibration, soundstage is automatically stable.
2.86 is odd one. little-bit dirty, but stage remains stable.

yes, last versions have something like as bug in code, used some #pragma, and dllimport maybe not optimized for data type of functions parameters. sound a bit terrible now.