lekt player

[jesuscheung]

i see v2.98.14xx, v2.98.15xx balance much better => sound haven't effect as 3D, intense or false emotion.

yes, since 2.98.14, 3D/bloated is gone.
i think because it extracts weak background/environment sound. (not sound from music)

v2.98.12xx a bit imbalance, listening piano i feel sop have little little effect, its sound have little shadow from left side/channel.
enjoing piano now with v2.98.15 256 ucx5, good balance, true sound. almost all things better than v2.98.14 160 xx, excluding scale of musical notes. ucx1 vibration better.
not sure, need more test.

i like both ucx1 c2 and cvx5
ucx1 c2's sound sinks more.
cvx5's sound rises more.

i notice for versions with notes sink more, the overall sound is automatically less dry more easy-going.

[jesuscheung]

ok... still prefer 2.98.14 ctlx 1 c2

[jesuscheung]

soundstage ucx5 is better.

ucx1 c1 and c2 soundstage is incorrect. sound size too big? stage too big?

[lekt]

soundstage ucx5 is better.

ucx1 c1 and c2 soundstage is incorrect. sound size too big? stage too big?

big but i think not "too". more vibration, more bass, stage size like as big. here's good focus. need make them more balance and they can be best.

i like organic sound of v2.98.14 160 ctlx1 c2. will back to its code and improve vibration, still a bit dirty vibrates. small buffer size seems lesser latency, musical notes more correct.
---------------------
average value of ucx5 maybe highest...

[jesuscheung]

soundstage ucx5 is better.

ucx1 c1 and c2 soundstage is incorrect. sound size too big? stage too big?

big but i think not "too". more vibration, more bass, stage size like as big. here's good focus. need make them more balance and they can be best.

yes, i totally misjudged. soundstage ok. sound is just forward.

a b compare ucx1 c2 256 vs ctlx1 c2 160 10 times with woman in love

ucx1 gives better texture quality.
256 gives typical 'harsher' sound.
160 refines the sound.
ctlx1 is less clearly. but very possible OS/hardware is muddy.

not easy to compare. coz buffer size is different.

[goon-heaven]

i feel buffer size 160 always more organic, more correct musical notes. but hard with code so then difficult improve vibration, will try later use its potential. now try 256 with more vibration.

uploaded:
v2.98.15 256 ucx1 c1
v2.98.15 256 ucx1 c2
v2.98.15 256 ucx5

optimized code from v2.98.12 256 sop and changed wasapi setting.
ucx1 used setting as v2.98.14 160 ctlx1, ucx5 as ctlx5.

JC, try with piano sound. how difficult with classical piano, i need try more, more...

v2.98.12 sop is more engaging and presense than v2.98.15 256 ucx1 c2.
RME HDSpe buffer range include 128 and 256, not 160 - v2.98.14 160 ctlx1 has less vibration.

[jesuscheung]

i googled your card RME HDSpe directly supports 128=3ms!

but last time min buffer showed up as 133(=3ms). so weird.

anyway, 256 and 160 share this clockrate 87075. try try...


-----

for 160, i guess you should select buffer = 64 (as supported by your card)

because 64*10/4=160... just guessing

[jesuscheung]

v2.98.12 sop is more engaging and presense than v2.98.15 256 ucx1 c2....

agree with you sop maybe the most engaging of all versions.
that's the sound musician hear in their head while they are playing an instrument.

[lekt]

all v2.98.xx 256 after 98.12 from one base code of v2.98.12 256 SHR. each version have private sound. retested, sop is most easy sound movement. seems here's correct setting. other versions improved some quality but changed this movement, sound a bit goes down and feel like weighty.

[goon-heaven]

i googled your card RME HDSpe directly supports 128=3ms!

but last time min buffer showed up as 133(=3ms). so weird.

anyway, 256 and 160 share this clockrate 87075. try try...


-----

for 160, i guess you should select buffer = 64 (as supported by your card)

because 64*10/4=160... just guessing

The HDSPe will easily play JPlay Kernel Streaming DirectStream into buffer 32, so I assume min buffer 133 is a result of all the clod in Wasapi layers blocking the sound on its way down to the KS layers?

I'll have a play with buffer sizes and clockrates..