Visual studio 2012 c++ and wasapi minimalist player

[goon-heaven]

Sorry, 2.59 rather too sophisicated, a little too sterile and uninvolving for me.

Gone back to 2.58 sse4 intel tot align - more alive and real.

[LowOrbit]

I have just spend my lunch break listening to 2.59. It's very good, possibly an improvement on other recent versions.

 

I am afraid I don't understand much of the feedback that gets posted here. As I understand it (and I do understand it, though am not a programmer) the software either plays the samples at the right time or it doesn't. It seems that the differing loop settings affect this timing at the clock level, and differences can be accounted for in the way this stream is decoded at the DAC. Get it right and the sound is clear with lots of detail and tonal accuracy (not the same as balance), get it wrong and details get smeared, the tonal balance gets distorted.

 

The only real attribute of the player that can be tweaked therefore is accuracy of reading the samples. The timing of samples reaching the dac is the only thing that can be influenced (assuming that the player is not performing any mathematical function on the data contained within the samples - and with MQn there is nothing of this nature occuring) and that is what causes changes in detail levels and tonal balance differences in what we hear.

 

Emotion is a meaningless metric - it is entirely subjective and means something different to each listener.

 

My own view is that you can only aim for the greatest accuracy of timing the data fed to the dac and all else will flow from that. Otherwise you are engaged in chasing "your" ideal sound and not necessarily what is in the file. The aim is to recreate as closely as possible what the mastering engineer intended. Tonal balance will vary from track to track, some songs will offer more engagement than others, but the player should only be delivering the samples. As sbgk has amply demonstrated that is not an easy ask, give the improvements evident in version after version over the last couple of weeks (let alone the earlier phase of development).

 

For me, emotional engagement with the music is dependent on many factors (mostly how I am feeling myself) and I am wary of anyone trying to "engineer" emotion into the replay mechanism. Get the accuracy right and when you're in the right frame of mind, the emotional connection will be there (as intended by the musicians/composer/producer/mastering engineer).

 

Also, in virtually every system I have ever heard, bass has been fuller and more superficially "there" in poor setups with bad timing and a lumpy response. In well balanced systems, bass is incredibly tuneful and goes much deeper, but with generally less subjective volume. Those systems also project a better sense of flow, rhythm and timing. And I think that is what sbgk is heading for. I certainly hope so.

[jesuscheung]

....

Emotion is a meaningless metric - it is entirely subjective and means something different to each listener.

 

My own view is that you can only aim for the greatest accuracy of timing the data fed to the dac and all else will flow from that. Otherwise you are engaged in chasing "your" ideal sound and not necessarily what is in the file.

....

 

if you have tears in your eyes or heartfelt, that's measurable emotion.

 

the thing is that there doesn't exist a complete and sound hardware that measures quality of audio output. that's why we are testing with our ears.

 

if such a thing exist, all hifi hypocrites will disappear.

[jrling]

Couldn't have put it better myself.

 

The point I expressed before was - 'The aim is to recreate as closely as possible what the mastering engineer intended' or perhaps faithfully reproduce what the mastering engineer put on the CD.

 

Anything else is 'colouration' and 'wrong'.

 

Jonathan

[Clive]

LowOrbit - if the bits are bits argument were valid then I would agree that total accuracy would be the objective. As we can tell from the MQn experiments the bits are bits argument is flawed. The problem then is what is the best sound? This can only be subjectively judged as there is no technical spec of bit accuracy to aim at for the final solution (MQn is bit perfect, this is just a tablestake).

[LowOrbit]

JC - don't misunderstand. I love to be moved by music too. I just don't think it can be engineered into the playback engine and your reaction to a certain piece may be very different to mine (and mine will vary depending on how tired or distracted I may be on a given day).

 

I also know we have to use our ears. That is one of the key learnings from sbgk's work. (You only have to listen to the top line systems by two different manufacturers (say dCS and Chord Electronics) to hear how different engineers interpret "best achievable" sound. And I agree that "most accurate" does not necessarily mean "best" because one of those is subjective and people interpret sound (especially tonal balance) differently. My own feeling is aim for most accurate and accept what comes out of the speakers (or headphones). That's what high fidelity means - and I agree we may not individually, subjectively like what we hear (certainly not every album, because they are not mixed to a common ideal).

[jesuscheung]

JC - don't misunderstand. I love to be moved by music too. I just don't think it can be engineered into the playback engine and your reaction to a certain piece may be very different to mine (and mine will vary depending on how tired or distracted I may be on a given day).

 

I also know we have to use our ears. That is one of the key learnings from sbgk's work. (You only have to listen to the top line systems by two different manufacturers (say dCS and Chord Electronics) to hear how different engineers interpret "best achievable" sound. And I agree that "most accurate" does not necessarily mean "best" because one of those is subjective and people interpret sound (especially tonal balance) differently. My own feeling is aim for most accurate and accept what comes out of the speakers (or headphones). That's what high fidelity means - and I agree we may not individually, subjectively like what we hear (certainly not every album, because they are not mixed to a common ideal).

 

you must have heard of the great Sennheiser Orpheus HE 90 headphones? the entire team of engineers are also musicians. musicians know best about true sound.

[erin]

2.53 eax does have good emotion/musicality. do you have a 650 or 701 etc headphone that smooths out the hardness?

 

I use speakers. My speakers are probably on the smoother side....

[lekt]

MQnPlay.exe 2.59 sse4 intel - alignment optimisation, add, dec

 

I wanted 2.58 sse4 intel to have a fuller less lean sound and think this has that, also sounds better overall. Had discarded the tot align version as didn't think it was that good, although very pleasant to listen to, this seems to have the best of both worlds.

 

be careful with 2.59 sse4 intel, great competition with 2.58 sse4 intel tot align !! i feel likely better (more strong, energy, power), i'm trying, need more listening. most important for me: vibration!

[jesuscheung]

be careful with 2.59 sse4 intel, great competition with 2.58 sse4 intel tot align !! i feel likely better (more strong, energy, power), i'm trying, need more listening. most important for me: vibration!

 

most important for me is emotion. we make a great team of testers.

[LowOrbit]

LowOrbit - if the bits are bits argument were valid then I would agree that total accuracy would be the objective. As we can tell from the MQn experiments the bits are bits argument is flawed. The problem then is what is the best sound? This can only be subjectively judged as there is no technical spec of bit accuracy to aim at for the final solution (MQn is bit perfect, this is just a tablestake).

 

Clive - I am not an advicate of the "bits is bits" argument because most of the proponents of that ignore the timing factor. The whole bits is just computer data is facile and meaningless and the argument deliberately ignores some of the key engineering problems and outcomes, which is why sbgk is being so successful, because he's ripped up that argument whilst not trying to change the bits themselves.

 

Best IS most accurate by any meaningful measure - subjective judgements are by definition unreliable. And - there is a technical spec of what accuracy constitutes. It's the definition in the Red Book Spec. It's just very difficult to achieve in playback, most approaches have failed. sgbk's getting closer from what I hear than most.

 

My point is fatuous to an extent because, as you all say, at the end of the day we are making a judgement with our ears. Few, if any of us, will ever know if 100% accuracy has been achieved. But if the developer isn't aiming for that, what is he aiming for? Some personal ideal that is a compromise and may not coincide with anyone else's interpretation.

 

It's a fascinating debate, probably a bit circular and I don't want to be divisive or dismiss anyone's opinion. Just trying to take a rational look at the process and understand what feedback really helps. I'm sure the developer will follow his chosen path to his own satisfaction and see if any of us are still here! I think we will be because I think he's a long way down the right track.

[erin]

I have a Celeron G530 which has 2 cores. Am I meant to be using mqncontrol.exe 2 core affinity, or something else?

Thank you

[LowOrbit]

you must have heard of the great Sennheiser Orpheus HE 90 headphones? the entire team of engineers are also musicians. musicians know best about true sound.

 

My own experience with musicians - most of them don't give a damn about hifi. They love the sound of the real.

 

If you've ever stood in front of a Marshall stack whilst playing a guitar, you'll know how far away from that experience we really are with our domestic replay systems.

[sbgk]

Sorry, 2.59 rather too sophisicated, a little too sterile and uninvolving for me.

Gone back to 2.58 sse4 intel tot align - more alive and real.

 

uploaded MQnPlay.exe 2.59 sse4 intel 8 8 and 8 4 alignment, I had neutered 2.59 sse4 intel a bit, see what you think about these. 2.59 would have been 4 4 if following this naming convention, see if you think it's got it's balls back

am listening to 8 8 at the moment, nice

[sbgk]

I have a Celeron G530 which has 2 cores. Am I meant to be using mqncontrol.exe 2 core affinity, or something else?

Thank you

 

don't think it matters for 2 core, best to use the latest version of 2 core affinity anyway

[goon-heaven]

uploaded MQnPlay.exe 2.59 sse4 intel 8 8 and 8 4 alignment, I had neutered 2.59 sse4 intel a bit, see what you think about these. 2.59 would have been 4 4 if following this naming convention, see if you think it's got it's balls back

am listening to 8 8 at the moment, nice

Thanks for putting the balls back! The music grabs me once more... Just have to remember to keep on breathing.

[erin]

JC - don't misunderstand. I love to be moved by music too. I just don't think it can be engineered into the playback engine and your reaction to a certain piece may be very different to mine (and mine will vary depending on how tired or distracted I may be on a given day).

 

I disagree with your comment. Like I wrote earlier, and speaking for myself, different versions of MQn can sound sterile and uninvolving, and by simply swapping versions of MQnplay, a dramatically different (involving and musical) sound can be had. So, I do think it can be engineered into the playback engine.

 

I also know we have to use our ears. That is one of the key learnings from sbgk's work. (You only have to listen to the top line systems by two different manufacturers (say dCS and Chord Electronics) to hear how different engineers interpret "best achievable" sound.

 

I think you might find that what high end manufacturers call "best achievable", if that is what they call it? Is actually mostly marketing hype. Or something that you have inferred or assumed that they have said. I've heard many top dollar components and left with a headache. Didn't like them at all!

But I take your point that different manufacturers gear does sound different.

 

And I agree that "most accurate" does not necessarily mean "best" because one of those is subjective and people interpret sound (especially tonal balance) differently. My own feeling is aim for most accurate and accept what comes out of the speakers (or headphones). That's what high fidelity means - and I agree we may not individually, subjectively like what we hear (certainly not every album, because they are not mixed to a common ideal).

 

True accuracy would be listening to the album played back in the mastering studio on the studio monitors it was mastered on.

Anything else is compromised. I've also heard so many stories of Pop / Rock albums that were mixed to sound good on Auratone speakers, intended to replicate car stereo and TV speakers. But I wouldn't want to listen back on those things.

 

Most speakers have around 4% distortion anyway. Accuracy is a myth.

I prefer a sound which is full range, as detailed as can be, whilst still maintaining musicality.

 

I've listened to far too many highly detailed, clinical, overly bright sounding systems and they do absolutely nothing for me.

 

The only question to ask yourself is - does my system sound good to me? Hopefully the answer is yes

[erin]

don't think it matters for 2 core, best to use the latest version of 2 core affinity anyway

 

 

Thanks. Its getting late here. I'll try it tomorrow and report back.

[sbgk]

I have just spend my lunch break listening to 2.59. It's very good, possibly an improvement on other recent versions.

 

I am afraid I don't understand much of the feedback that gets posted here. As I understand it (and I do understand it, though am not a programmer) the software either plays the samples at the right time or it doesn't. It seems that the differing loop settings affect this timing at the clock level, and differences can be accounted for in the way this stream is decoded at the DAC. Get it right and the sound is clear with lots of detail and tonal accuracy (not the same as balance), get it wrong and details get smeared, the tonal balance gets distorted.

 

The only real attribute of the player that can be tweaked therefore is accuracy of reading the samples. The timing of samples reaching the dac is the only thing that can be influenced (assuming that the player is not performing any mathematical function on the data contained within the samples - and with MQn there is nothing of this nature occuring) and that is what causes changes in detail levels and tonal balance differences in what we hear.

 

as I have learned about the what the computer has to do to deliver those bits the more I realise just how much the sound is affected by everything going on in the cpu

 

read manual 2 at the following link to understand a bit more about how the cpu works Software optimization resources. C++ and assembly. Windows, Linux, BSD, Mac OS X

 

the alignment settings I have been trying recently are about the fact the instructions are loaded more quickly by the cpu if they are aligned on certain memory boundaries, so when I say 8 4 that means the entry point to the function is aligned on 8 bytes and exit on 4, so you can hear the effect this has on the sound as 2.59 used 4 4 and people didn't like it, I guess the cpu had to do some extra work because the instructions weren't loaded optimally.

 

The problem with the bits are bits argument is that no one has been able to prove by measurement that the sound is the same or different. Until it can be measured we might as well say there are noise particles produced at various stages of the chain and the aim is to stop the production of these noise particles from polluting the music stream, it may not be real, but at least it reflects what we hear.

[sbgk]

True accuracy would be listening to the album played back in the mastering studio on the studio monitors it was mastered on.

 

My ATC speakers were bought second hand and when I phoned ATC to enquire about upgrading the drive units they said do you realise who they belonged to before ? turns out they were made for Jim Guthrie - sound engineer for Pink Floyd amongst others, so that's my claim to listening to music that's close to the mastering studio, Dark Side of the Moon/Wish you were here sounds sweet on them. They have yew veneer so probably were used domestically, but even so.

[goon-heaven]

read manual 2 at the following link to understand a bit more about how the cpu works Software optimization resources. C++ and assembly. Windows, Linux, BSD, Mac OS X

 

the alignment settings I have been trying recently are about the fact the instructions are loaded more quickly by the cpu if they are aligned on certain memory boundaries, so when I say 8 4 that means the entry point to the function is aligned on 8 bytes and exit on 4, so you can hear the effect this has on the sound as 2.59 used 4 4 and people didn't like it, I guess the cpu had to do some extra work because the instructions weren't loaded optimally.

 

Thanks for explaining what you are tinkering with down there.

The 8 8 results of your delving is sure pulling the strings of this puppet.. grabbing my ears, wagging my head and tapping my feet!

[jesuscheung]

 

My ATC speakers were bought second hand and when I phoned ATC to enquire about upgrading the drive units they said do you realise who they belonged to before ? turns out they were made for Jim Guthrie - sound engineer for Pink Floyd amongst others, so that's my claim to listening to music that's close to the mastering studio, Dark Side of the Moon/Wish you were here sounds sweet on them. They have yew veneer so probably were used domestically, but even so.

 

lucky you! you have the exact speakers needed for Pink Floyd. zero distortion.

[erin]

 

My ATC speakers were bought second hand and when I phoned ATC to enquire about upgrading the drive units they said do you realise who they belonged to before ? turns out they were made for Jim Guthrie - sound engineer for Pink Floyd amongst others, so that's my claim to listening to music that's close to the mastering studio, Dark Side of the Moon/Wish you were here sounds sweet on them. They have yew veneer so probably were used domestically, but even so.

 

Perfect!

[tonerei]

I came in to this testing late and the first version that really rocked my boat was the 2.53 sse4 as mentioned by Goon-Heaven above.

Am struggling to find enough time to compare all the versions that are coming out but have condensed my trials to the 'best' recommended ones here. Has it reached a point where it is now people's preferences that are dictating the reactions to the myriad versions or is it still improving? I do realise I need to go through the thread and start reading the technical bits.

 

One point for me anyway the music I am playing from 2.53 on regardless of type has great detail,clear,concise not sharp or overbearing. Vocals are sublime even my favourite singers who might not necessarily be the most in tune performers on the planet.Bass is tuneful and soundstage is deep. On really good vocalists the guy or gal seems to be literally touchable just in front of you.

 

Thank god for Sbgk

[jesuscheung]

be careful with 2.59 sse4 intel, great competition with 2.58 sse4 intel tot align !! i feel likely better (more strong, energy, power), i'm trying, need more listening. most important for me: vibration!

 

 

 

2.59 8 8 vs 2.59 8 4 vs 2.58 sse4 intel tot align

-tot align is more real. fuller. deeper.

-2.59 is cleaner.

-8 8 and 4 8 differs mainly in stage? both good. i might prefer 4 8.

-think 2.59 is able to "zoom in & out". 3D effect? feels more lively. might cause fatigue.

-tot align has better vocal. more emotion.

 

is it possible to make a flat stage like rax no zi? it is simple and musical.