Tír Na HiFi

jesuscheung wrote:i see lekt.exe=71KB.

that sounds like thousands lines of code...

not that is, JC. lekt's code very short, ~8KB. as you see many versions 8-9KB only. that's used method without maping data embedded into file.
with last current method if biger buffer size then biger file size, but need keep code simple and shortly. also depend what mode chooses for compiler, if compiler optimize for fast code, for small size and remove all metadata then very small file size. if lets all as is then very big size (~ x 4) but sound much better. it's secret, in my case, hehe... lekt's code may include in only one main function, as many versions playing good sound.

see v2.83 R1 sound good, need a bit improve balance. prefer it.

lekt wrote:

jesuscheung wrote:i see lekt.exe=71KB.

that sounds like thousands lines of code...

not that is, JC. lekt's code very short, ~8KB. as you see many versions 8-9KB only. that's used method without maping data embedded into file.
with last current method if biger buffer size then biger file size, but need keep code simple and shortly. also depend what mode chooses for compiler, if compiler optimize for fast code, for small size and remove all metadata then very small file size. if lets all as is then very big size (~ x 4) but sound much better. it's secret, in my case, hehe... lekt's code may include in only one main function, as many versions playing good sound.

see v2.83 R1 sound good, need a bit improve balance. prefer it.

what's secret about copying MQn ? You're a funny guy.

lekt wrote: think almost C functions give more rich and density, cleaness also, than C++ . need improve vibration so make sound more soft, ....

agree! now that sound is cleaner, it can fit-in more vibrations

sbgk wrote: what's secret about copying MQn ? You're a funny guy.

provocation, dangerous. not good.

jesuscheung wrote:agree! now that sound is cleaner, it can fit-in more vibrations

uploaded v2.84 1024

vibration much improved, good balance. i think sound less intense, no strain. used new method that lets make small file size for larger buffer size 1024. this version's file size only 10KB.
try it and enjoy.

lekt wrote:

jesuscheung wrote:agree! now that sound is cleaner, it can fit-in more vibrations

uploaded v2.84 1024

vibration much improved, good balance. i think sound less intense, no strain. used new method that lets make small file size for larger buffer size 1024. this version's file size only 10KB.
try it and enjoy.

very rich sound! transparent/air just enough.
all things kept, i believe.

something missing. not head banging!

EDIT
made a bad morning judgement. this is definitely new best

i think the amount of vibration has doubled in 2.84?!

all good thing kept. 2.84 is a treat listening to metal/rock!

actually, the soundstage is also new best here.
i can hear things close to me, further from me. positioning is natural. all instruments work well together.

gonna need 352, 320 versions for 2.84, if possible...

jesuscheung wrote:i think the amount of vibration has doubled in 2.84?!

all good thing kept. 2.84 is a treat listening to metal/rock!

actually, the soundstage is also new best here.
i can hear things close to me, further from me. positioning is natural. all instruments work well together.

gonna need 352, 320 versions for 2.84, if possible...

v2.84 1024 use multi threading (7 threads running for render loop, interesting result. only code not optimized, using shared memory variable, sound have shadow, need more sharpness and clarity. will go back to it, 7 threads seems enough, latency maybe ~0, but code is complex, maybe exist bugs, will review. it balance is best, so soundstage is best!

now uploaded v2.84.1 1024

this version uses 3 threads, shared memory lesser affect, used big file size method. sound clarity, very good.
will optimize code of two these versions, love multithreading now, hehe...
need larger buffer size, JC, my laptop onboard sound card can play only 1024 max

and uploaded v2.84.2 1024

more vibration, sound very rich.
OK, now will review and back to v2.84 code, its sound have potential to hifi.

lekt wrote:......
v2.84 1024 use multi threading (7 threads running for render loop, interesting result. only code not optimized, using shared memory variable, sound have shadow, need more sharpness and clarity. will go back to it, 7 threads seems enough, latency maybe ~0, but code is complex, maybe exist bugs, will review. it balance is best, so soundstage is best!
....

2.84 > 2.84.1
2.84 > 2.84.2
much better... hehe

2.84.2's vibration is 'raw', not refined- can create earaches.

2.84 only weaknesses are -0.5% energy and -0.5% brightness. all things balance.

2.84 maybe difficult to 'drive'. i have an older windows backup. about 5-10% more dirty than my current windows. in this older windows, 2.84 is too 'heavy', i.e. too much noise and vibration mixing together.

in my current windows, 2.84 is god-like. can still be better. i still feel dirty because of OS.