Visual studio 2012 c++ and wasapi minimalist player

[sbgk]

None of mqn versions work with 16 bit files (tried 1.10; 2.35..2.40; 3.10). I always get same info in cmd window:

"valid bits 16 conteiner bits 24 rate 44100

Audio device does not support the requested format."

And then it force closes.

Hope you could develop a version of mqncontrol, capable to "talk" to my dac for also playing 16/44 and 24/48 (now it shows as everything would be alright, but no sound and just hungs and i can not even kill it and have to make reset). Maybe this version would work for others who struggle like me. Huge thanks in advance!!!

 

note: i do not have jplay installed in my fresh 2012 server gui install yet, so i guess system timer is untouched.

 

have uploaded mqncontrol.exe 2496 1644 24 bit v6 into test for you to try 16 bit

[jrling]

Hi Jonathan,

 

He can't and shouldn't.

That's why I wrote 'I would like too encourage you too follow your own ears.'

 

I disable all Performance Counters with exctrist.exe

 

Please report back if you hear any improvement in SQ when disabling performance counters and lowering USB polling frequency.

The SQ improvement in XP is significant

Especially lowering USB polling frequency

But in 2012 is only subtle

I could be fooled by the placebo-effect.

 

Thanks Mark - will try the tweaks and report back. I recall the USB polling now from my CMP/XP days which thankfully are long gone - for me plain vanilla MQn/WS2012 GUI wipes the floor with CMP even tweaked to the extreme.

 

WS2012/MQn out of the box is so good, a testament to SBGK's excellent approach and coding, that such tweaks can have very minimal effect and therefore difficult to detect. WS2012 is also a v good OS for Windows-based audio. But I always leave them in place if they don't degrade SQ as there is probably a compounding effect of loads of small improvements.

 

I tell SBGK what my ears tell me always!

 

Jonathan

[jesuscheung]

have uploaded mqncontrol.exe 2496 1644 24 bit v6 into test for you to try 16 bit

 

"valid bits 24 container bits 24 rate 96000

We ended up with a period of 107500 hns or 1032 frames."

[jesuscheung]

Hi sbgk

....Reading comments like "if you use good recording, more difficult to tell differences in piano sound." makes me wonder. Big chance that this way you are 'voicing' MNq too compensate for short comings in lesser recordings or equipment being used. Not too mention the acquired sound-taste of the reviewer.....

 

i am trying to say is that MQn is able extract detail/carity from bad piano recordings that other players cannot. in good recordings, music is more easy to present, most players do a good job- so can take longer for me to make judgement. sorry for my bad english.

[jesuscheung]

I used this tweak in my cMP / cPlay setup

 

Step 7 - Reduce USB Polling

 

Default polling of USB (Controller) devices under XP is 1ms. Use following registry change for each USB Controller to increase polling interval to 5ms.

Select HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\Class\{36FC9E60-C465-11CF-8056-444553540000}\0000

Right-click in right pane > Select New > DWORD > Rename to 'IdleEnable' > Set value to 1

Repeat for each subkey \0000, \0001 ... that has the 'Controller' key

 

I found the same USB polling registry values in server 2012

 

Same for the exctrlst.exe application

Two-Minute Drill: Disabled performance counters and Exctrlst.exe - Ask the Performance Team - Site Home - TechNet Blogs

 

May be I'm fooled by the placebo-effect, but I think I also hear sound quality improvement not only in XP but also in server 2012.

 

thx for the tweak! works for me. i used to just disable the kernel driver for performance counters, but there were side-effects. this method is better!

[Robertas Morkunas]

have uploaded mqncontrol.exe 2496 1644 24 bit v6 into test for you to try 16 bit

 

nope, it still doesn't make a sound and same error message though 24/96 still works

sorry, for pushing you so hard..

[jesuscheung]

I found 2.41 piano to be quite different to 2.40, shall see what it sounds like today. I guess music is mixed using studio monitors so it is likely that a good player will make it sound like that, unless there is a domestic listening button on the mixing desk which they press just before release. If you are at a live concert all the instruments are heard and you just have to concentrate on the instrument you are interested in or just let the whole thing fuse together. Think 2.40/2.41 are a step above the others in that they can resolve detail and staging, I find them more entertaining.

 

yes. up to now 2.40 is the most correct of all versions.

 

in 2.40 there is an very explicit indication of where every instrument is positioned- it creates an separation effect of sound, how can you be sure the music composer wasn't intended for any two instruments to integrate as one?

 

in 2.40 every instrument sounds important. yes, it is good if main music is not what you are after to listen. this creates a problem for listener to make an effort to find, locate and then focus on the main music. i mean, any instrument can be the main music in 2.40

 

that's why i call 2.40 a hd800 headphone. as great as hd800 is in clarity/detail, it will never be a great musical headphone.

 

2.37 wc no xmm sub -128 noaff is a good version for effortless listening in which main music is the focus, background music is just there to bring out the soul.

 

2.41 is like 2.40. yes they are different. same conclusion.

[jesuscheung]

sbgk, you should really consider that two instruments cannot be physically place at the same place, that's why they are positioned differently in space. but the music composer may be intended for an integration of sound for the listener. if you separate the sound, the result can be against the original intention.

[jrling]

sbgk, you should really consider that two instruments cannot be physically place at the same place, that's why they are positioned differently in space. but the music composer may be intended for an integration of sound for the listener. if you separate the sound, the result can be against the original intention.

 

I think the truth is that we can only hear (at best) to whatever the producer and the sound engineer decide is the way that they want to mix the tracks. IMHO many of them don't take trouble or pay attention to what you say or just don't really care or understand.

 

So SBGK is surely aiming with MQn to recover as close to exactly what is on the very processed mix on the CD as best he can.

 

Or put it another way - "Rubbish in, rubbish out."

[sbgk]

nope, it still doesn't make a sound and same error message though 24/96 still works

sorry, for pushing you so hard..

 

from your wasapi_test results it looks like 16 bit is not an option 32000 24 2 192000 6 65534 22 24 3 - Format works ok

44100 24 2 264600 6 65534 22 24 3 - Format works ok

48000 24 2 288000 6 65534 22 24 3 - Format works ok

88200 24 2 529200 6 65534 22 24 3 - Format works ok

96000 24 2 576000 6 65534 22 24 3 - Format works ok is it possible to configure your dac to be 16 bit and 24 bit ?

[sbgk]

sbgk, you should really consider that two instruments cannot be physically place at the same place, that's why they are positioned differently in space. but the music composer may be intended for an integration of sound for the listener. if you separate the sound, the result can be against the original intention.

I've never heard of this concept, do you have an example ? All I am doing is trying to get the signal through the cpu as efficiently as possible and measuring it by the ability to make a stereo image as close to reality as possible, I guess we need to take a consensus view of what the best version is. Don't think there is a best version yet, I can see significant gains by putting the complete render loop in assembly, which is what I'll try and do this week.

[jesuscheung]

I've never heard of this concept, do you have an example ? All I am doing is trying to get the signal through the cpu as efficiently as possible and measuring it by the ability to make a stereo image as close to reality as possible, I guess we need to take a consensus view of what the best version is. Don't think there is a best version yet, I can see significant gains by putting the complete render loop in assembly, which is what I'll try and do this week.

 

hmm... forget what i say. my explanation is confusing. just make a better version.

[Robertas Morkunas]

from your wasapi_test results it looks like 16 bit is not an option 32000 24 2 192000 6 65534 22 24 3 - Format works ok

44100 24 2 264600 6 65534 22 24 3 - Format works ok

48000 24 2 288000 6 65534 22 24 3 - Format works ok

88200 24 2 529200 6 65534 22 24 3 - Format works ok

96000 24 2 576000 6 65534 22 24 3 - Format works ok is it possible to configure your dac to be 16 bit and 24 bit ?

 

hmz.. then how can i play everything in foobar (with Kernel streaming and with Wasapi: both after setting 24bit in foobar's output settings page) and in Jplay the same ?

I can not configure dac as it has native drivers (i do not need to install anything separately) and there are no settings to select, just in the players itselves. Any advice?

[nige2000]

24/88 and 24/192

 

would be nice

 

any ram raming advice jesuscheung?

[jrling]

I used this tweak in my cMP / cPlay setup

 

Step 7 - Reduce USB Polling

 

Default polling of USB (Controller) devices under XP is 1ms. Use following registry change for each USB Controller to increase polling interval to 5ms.

Select HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\Class\{36FC9E60-C465-11CF-8056-444553540000}\0000

Right-click in right pane > Select New > DWORD > Rename to 'IdleEnable' > Set value to 1

Repeat for each subkey \0000, \0001 ... that has the 'Controller' key

 

I found the same USB polling registry values in server 2012

 

Same for the exctrlst.exe application

Two-Minute Drill: Disabled performance counters and Exctrlst.exe - Ask the Performance Team - Site Home - TechNet Blogs

 

May be I'm fooled by the placebo-effect, but I think I also hear sound quality improvement not only in XP but also in server 2012.

 

Hi Mark - just checking one point.

Regedit says that DWORD is for 32 bit and QWORD is for 64 bit.

Is it DWORD to edit if one is using WS2012 64 bit?

 

Thanks

Jonathan

[sbgk]

hmz.. then how can i play everything in foobar (with Kernel streaming and with Wasapi: both after setting 24bit in foobar's output settings page) and in Jplay the same ?

I can not configure dac as it has native drivers (i do not need to install anything separately) and there are no settings to select, just in the players itselves. Any advice?

 

they are converting 16 to 24 bit, well adding a zero byte in to make it 24.

 

I am converting the 16 bit to 24 bit in the test version, must be doing it incorrectly

[sbgk]

24/88 and 24/192

 

would be nice

 

any ram raming advice jesuscheung?

 

shall do it this week hopefully, slightly complicated because it won't align with 8192 Bytes, so lots of changes.

[nige2000]

[h=1]ram timings i meant[/h]what works good

[sbgk]

hmm... forget what i say. my explanation is confusing. just make a better version.

 

uploaded 2.42 , better than 2.40 and 2.41

 

also uploaded 2.41 movq which uses 64 bit registers, sounds pretty bad, this is what your old pentium would sound like.

[Robertas Morkunas]

they are converting 16 to 24 bit, well adding a zero byte in to make it 24.

 

I am converting the 16 bit to 24 bit in the test version, must be doing it incorrectly

 

Understood. Ok, take your time. No need to rush anywhere with it. Thanks for all your efforts.

[lekt]

I've never heard of this concept, do you have an example ? All I am doing is trying to get the signal through the cpu as efficiently as possible and measuring it by the ability to make a stereo image as close to reality as possible, I guess we need to take a consensus view of what the best version is. Don't think there is a best version yet, I can see significant gains by putting the complete render loop in assembly, which is what I'll try and do this week.

 

talk about the technical criteria (non musical) of 2.40: dimension (x, y, z) and weight of sound are incorrect a bit ("3D is an issue for me with piano", as jesuscheung said), then i feel tired when long time listening. nopgomemcpy does not get tired, though listening all day. other criteria is OK (SQ, detail/clearly/clarity...)

vocal is pretty but near at us => clarity in over norm, and sadness reduced.

SQ is better than nopgomemcpy, but i still use nopgomemcpy because its music is more sad and i feel not tired when long playing.

i think nopgomemcpy is absolute correct in the technical criteria, on my system frequently sound resonance comes after ~30 min playing (sound so much better when resonance comes). there maybe other causes and i am probably not right.

[sbgk]

ah, but your error message said 16 in 24 format not supported, so maybe needs actual conversion, suppose you could do it in sox using the same method as the flac conversion.

[sbgk]

talk about the technical criteria (non musical) of 2.40: dimension (x, y, z) and weight of sound are incorrect a bit ("3D is an issue for me with piano", as jesuscheung said), then i feel tired when long time listening. nopgomemcpy does not get tired, though listening all day. other criteria is OK (SQ, detail/clearly/clarity...)

vocal is pretty but near at us => clarity in over norm, and sadness reduced.

SQ is better than nopgomemcpy, but i still use nopgomemcpy because its music is more sad and i feel not tired when long playing.

i think nopgomemcpy is absolute correct in the technical criteria, on my system frequently sound resonance comes after ~30 min playing (sound so much better when resonance comes). there maybe other causes and i am probably not right.

 

try 2.42. I find nopgomemcpy to be too digital sounding and tiresome after 30 seconds. If you want sad music just play Coldplay.

 

you guys complaining about 3d, are you aware that modern cd production makes the sound more like it is coming out of the speakers (don't know what they call it, but same as they make the volume louder and compress the sound), the Eminem track - Stan I tried the other night had this effect to a great extent and I just dismissed it as that's how it was produced, it's not an effect of the player, it's just crap production.

[jrling]

try 2.42. I find nopgomemcpy to be too digital sounding and tiresome after 30 seconds. If you want sad music just play Coldplay.

 

you guys complaining about 3d, are you aware that modern cd production makes the sound more like it is coming out of the speakers (don't know what they call it, but same as they make the volume louder and compress the sound), the Eminem track - Stan I tried the other night had this effect to a great extent and I just dismissed it as that's how it was produced, it's not an effect of the player, it's just crap production.

 

My point exactly. Production values these days are dropping. 'The Loudness Wars' are the current preoccupation for labels.

 

Have not heard 2.42, but currently 2.41 rules for me. It has everything. Detail but warmth as well, lovely 3D staging ....

 

Ridiculously and scarily good.

 

Well done SBGK.

[sbgk]

uploaded 2.42 , better than 2.40 and 2.41

 

also uploaded 2.41 movq which uses 64 bit registers, sounds pretty bad, this is what your old pentium would sound like.

 

uploaded 2.42 sse4 intel v2 for comparison with 2.42 sse4 intel, be interested to know what people think