Maybe this could help:
 
 
V-Plugs has released Mr Soundman, a challenging free ear training game for music producers and sound engineers. Mr. Soundman is a powerful tool for assessing and improving frequency detection, a critical skill for mixing, recording and producing music. It also enables users to win excellence awards and prizes, and to participate in Mr. Soundman special competitions.
 

 
 
The Game
Mr. Soundman behaves a bit like an arcade game with an intuitive GUI, and rapidly increasing difficulty. The basic concept is simple: Mr. Soundman plays an audio source with a noticeable boost somewhere across the frequency range. The mission is to identify the boosted frequency.
The closer the answer is to the boosted frequency the more points the user gains. As the game progresses the 'right answer range' becomes narrower, so users have to provide more precise answers. The game starts with a wide 'right answer range' of 3 octaves, but only the best will get to the ¼ octave zone.
 
Prizes, Awards and Competitions
Depending on their score, users can win three excellence awards: the Bronze Ear, the Silver Ear and the Golden Ear award. Each of these awards grants a special prize. The ambitious user can also participate in Mr. Soundman's special competitions that will be announced from time to time, and compete with colleagues and friends around the globe.
 
Mr. Soundman Premium
The premium version of Mr. Soundman gives an advanced ear training solution for those who want to take audio ear training a step forward. It offers more sound sources (and sound categories), focused training sessions, and even a hearing test. It also enable the user to load his own sounds, to build and manage his personal practice folder, and to work on and improve certain aspects of his listening skills.
 
Pricing & Availability
Mr. Soundman is available as a standalone application or as a VST instrument for Windows. The basic version is free, and the premium version costs €29.
 
 
 

mike_mccue


Hi Danny,

I think we all agree it's just about impossible to say you can hear the difference between 64 and 32 point floating.

I am probably an outlier (Like Jim R) that can imagine up a way to appreciate the extra detail. Things like fades, and trails and stuff are exactly where one might find appreciation, if any, to the extra detail.

At some point I noted, and Drew seemed to agree, that any detail we are talking about in high bit depth files gets thrown away when you go to 16bit CDs etc.  That was sort of a "why bother" acknowledgement.


I pointed out that on my system using the extra detail consumes 1% of CPU on a system where I usually have 80% CPU left over and being wasted anyways. The reason this works so well is that Pentiums evolved to do native 64 bit processing and when that happened Cakewalk just programmed for it's use.


The questions about just listening have been answered by numerous independent studies that inform us that none of us seem to have trust worthy ears... no matter how hard we try to use them or how disciplined we think we are about listening. Time and time again, listening tests have been proven that results are almost always inconclusive.


The other thing is that each persons opinion will be informed by their personal experience. So, someone with monitors set up against the wall and a sub woofer off in the corner is going to have one experience and the person with a monitor setup that was setup with some acknowledgement of a respect for basic acoustics is going to have another experience. Discussions are rarely "like to like"... which undermines the usefulness of the discussion.

One thing is for sure, the only practical way to hear anything going on down at -126dBFS levels Drew is speaking of is to stand in front a huge PA stack and listen real carefully.


:-)

I am personally glad that the mix engines we are using are so capable that there is really nothing to worry about as long as your system is working smooth.

best regards,
mike




P.S. I am currently reading a book about the LHC over near Geneva. The scientists there are breaking atoms apart and finding info about stuff that, when I was a school boy, we were told didn't even exist.
That is why many of us that may not hear a difference with our ears choose to leave some wiggle room for the possibility that people that say they hear stuff we don't.... might actually sense what they think they sense.
For me it seems harmless to let the Pentium run the math in 64bit... just in case there's something special to be enjoyed in the cross fades and the reverb tails and the blending of the violin section that I don't yet recognize.

Hi Mike,
 
I think that's all I really wanted people to agree with to be honest. LOL! I mean seriously Mike...think of it from my point of view. I got this guy showing me this big math problem that he claims has errors in it. Yet, if I listen to the song that has these errors and then the same song without the errors, and they sound exactly the same, how would he expect me to buy into it? I've just always been a guy that will only use something or do something that makes that "wow" difference. Of course that all stems down to whether or not I'm happy with my production or not, ya know? Like for example, if I was not happy with my quality or whatever, I'm the first to tell you that I'll try everything and anything to gain fidelity. So I'd be trying this stuff like mad. But thank God, because I'm so content, something has to really win me over in order for me to believe in it. Make sense at all? I see your point too with "hey, it doesn't cost anything...these boxes are there, it doesn't mess with my system at all, ok, might as well leave them." That makes total sense. However, IF there is a way to really hear this to where it's drastic, I really want to learn how to. Just about all the stuff I've learned in this field was about "learning how to listen" when my ears couldn't pick out the right stuff.

Now see how you mentioned crossfades and reverb tails...blending etc, NOW we're getting somewhere. I would have never thought this float thing would have anything to do with stuff like that...so if that's the case, this intrigues me. The more of that little stuff that it could affect, the more it could be improving audio. We're not just listening for a difference in the song on the whole. See, that's what I always listened for....the song on the whole. Listen this way, ok sounds good...listen this way...ok sounds good here too! So I assumed there's no reason to use this stuff. But now that you throw stuff like that at me, it's quite obvious I was listening carefully, but not to stuff like that and this opens up a new world. Still nothing drastic, but at least I know what to listen for now. I think that's all I needed to know. Once you know what to listen for, then you can really make up your mind whether or not it's something you need or not I think. My system has always been like Bub's. A definite increase in cpu as well as clicks, pops and sputters. When I first met Noel, he pm'd me to try and find out why I was having this problem. I just could never get the 64 bit mix engine to work and never did the 32 bit thing because cards don't support it. I figured...I got a 24 bit card, it should do 24 like it was made for, ya know? But I'm going to revisit all this now with a new attention to detail.
 
On the box that Jim built for me, it works...but when I export audio, about every 3 songs or so, I get mad artifacts in the exported file. Actually, the last time I tried it, all I got was a noise file with no audio. This was me at 24/48 with the 64 bit mix engine enabled. Nothing but white noise and no audio. I went back to the project, exported again...same thing. So I turned off 64 bit mix, and exported again...there was my tune.
 
Did I need 32 bit float enabled? That's another question I got. In Sonar's audio data options, do you (and anyone else) ever use the 64 bit options there? Like, if I record at 24/48....what should I have the audio file options set for to really see how this works? Record bit and render....you don't put both of those on 64 do you? Currently I have the following options.
 
24/48, no 64 bit mix engine. In audio data options, 24, 24 and original.
 
Set me up with what would be considered the most perfect 24/48 options if you would? Maybe the combination I was using was messing everything up. I really want to try this stuff and come to grips with it. Thanks for the insight. :)
 
-Danny

we have come full circle! what has been sugested by some is when in 64fp mode we can see an increase in cpu usage because some plugs are enabled for 64fp mode and some arnt, vc64 channel strip anyone?

now as i suggested run everthing at 32fp and the system works better because everything within sonar is 32fp.

peace

SteveStrummerUK

Maybe this could help:
 
V-Plugs has released Mr Soundman ...
 

 
 
Mr. Soundman Premium
The premium version of Mr. Soundman gives an advanced ear training solution for those who want to take audio ear training a step forward. It offers more sound sources (and sound categories), focused training sessions, and even a hearing test. It also enable the user to load his own sounds, to build and manage his personal practice folder, and to work on and improve certain aspects of his listening skills.
 
Pricing & Availability
Mr. Soundman is available as a standalone application or as a VST instrument for Windows. The basic version is free, and the premium version costs €29.

I'm sorry Steve, this is just a childish attempt to sell a product. We all know this is garbage. We would appreciate you not trying to confuse the conversation here with your subliminal sales tactics.

I know that may be a tad harsh, but it's true.

I would think you could see beyond this childish behavior but apparently I'm wrong.

Bub.

Danny Danzi

drewfx1


Danny if the errors peak at something like -120dBFS with an average level 15dB below that, then they sound like silence.

But more seriously, the short answer is it should sound like noise that's been amplitude modulated by your actual signal's level in 6dB steps.


I would say if you're really interested you can do something similar to my tests:

Take a large project with as many tracks and busses as possible, make sure you turn off dither everywhere, and export it to two 24bit files using both the 32bit and 64bit engines. Then you can invert one and add them together to see what the difference signal is. 

You might have to add lots of gain to even hear it at all. Think in terms of +40dB, and maybe much, much more.

Also keep in mind that listening to a difference signal out of context can be misleading, as in the real world your actual signal will mask much or all of what's there. Especially for stuff like this where the error level goes down when your signal gets quieter.

Which is why a comparison of the audio files is always the best approach. But the difference signal can give you an idea what to listen for.

Drew, thanks for the reply. :) I guess maybe I don't understand this stuff at all and it's just way over my head. Your example still doesn't really help me though. Don't take that the wrong way...lol...I'll explain and I WILL try your test you mentioned. But let me just ask because this still seems to not make sense to me.
 
If I have to take these 2 files and invert one of them to hear a difference, what does that really tell me...there's a volume difference? Is this whole thing about additional loudness or something? Like, you see how you mentioned the test and then we inverted....can't I just listen to one file alone, then the other and see what differences I hear?


I think that is my whole point to my issue with this. Why do we have to do something else to hear something that may be wrong? Does that make sense at all and can you at lease see where I'm coming from? I'm honestly not trying to be stubborn with any of this. I'm trying to understand how if I can't hear a difference with something by just listening to it, how doing  the test is going to show me any different. So for me to truly hear how this works, I have to do the test and invert to hear differences, yet if I listen to both files individually, there won't be an audible difference. LOL! Can you see the method to my madness with this? If I'm totally missing something here, I'm all ears and like I said, I will try the test just for the sake of it. But again, we're using a "method" to hear these differences where on their own, we won't be able to tell. If that is indeed the case, is it really worth it to go through these tests other than to say "ok, one file sounds a bit different than the other...but only if I run them at the same time and invert.....I can't hear a major difference unless I use this test." See my point at all? LOL! Honest I'm not trying to give anyone a hard time or anything. I just doesn't make sense to me.  
 
-Danny

You are entirely correct, and in fact a direct comparison is the only real way to do a meaningful test.

The reasons for doing a null test, which results in a signal that is the differences between the two files (i.e. containing just the noise/distortion/artifacts/errors but not the original audio), is either to:

1. Determine whether there are any differences and what their nature (noise? distortion?  frequencies involved?) might be, and especially to objectively measure/analyze these differences. If you do measurements and find that the magnitude of the differences between 2 files is ridiculously small, then you safely can conclude the differences won't matter.

2. Just to get an idea of what the differences sounds like in isolation. You asked "What does it sound like?", so that's why I suggested this. But as I said earlier, it doesn't make sense to draw conclusions about audibility from listening to a difference signal in isolation since some or all of the noise/distortion/artifacts will be almost always be masked by your actual signal. But it can give you an idea for what to listen for when comparing the 2 actual signals to see if you hear any difference.

And if you find you have to amplify the difference signal by +40dB or more to hear it at all, even in isolation, you might not want to bother with the hassle of doing a test and straining to determine if you think you can hear any difference in the actual audio.

Mike, Danny

As Noel or one of the other knowlegable programmers explained when the 64-bit engine was introduced, it doesn't have anything to do with a 64-bit OS but is depth at which the math is done internally w/in SONAR.  the theory being as I understood it that the bigger numbers sued the less the rounding errors and the less likely the math will introduce errors and spit out something more, or less, than was put into it.  For well recorded and mixed material, there shouldn't be any difference between 32 and 64 bits, or 24 bits for that matter.  However, if you run every channel at max and red-light everything, the 64-bit engine would output less errors and hense nasty digitial distortion.  Much like bub's rogue sample that caused clicks (ie., short periods of distortion) at 24 bits but not when rendered at 32.  My analogy is a premium preamp that has more headroom than a cheap ic one and doesn't crap out when overdriven but smoothly shifts into overdrive.  Maybe Drew could explain it more mathmatically.  I don't plan to record or mix w/ overs, but use it as a safety feature.  Even on my old dual-core system I have at home the 64-bit engine doesn't seem to add much, if any, overhead.

@

Hi AT,

My DAW is using a 32bit OS with the 64bit double precision mix engine turned on.

I've been using it since the day it came out.

Back in the day, the Pentium chips introduced their 64bit processing and that's when Cakewalk was positioned as a leading partner with Intel and Microsoft as those two companies were making the instruction sets and APIs.

The 64 bit mix engine uses the Pentiums capabilities... and of course has nothing to do with the OS.

I was the guy who pointed out yesterday that turning it on seems to add, maybe, 1% CPU use on my system and that system usually has over 80% of it's resources left open.

Drew was showing us some math that addresses the scenario you are describing:

"However, if you run every channel at max and red-light everything, the 64-bit engine would output less errors and hense nasty digitial distortion."

He attempted to quantify what you are describing and he showed that even in 32 the amplitude level of any errors is very very very small.


 best regards,
mike

musicman100

Bub

And 3 ....

2 ...

1 ...

There she goes !!!!!

NOPE STILL UPSTAIRS....

I'm trying my best! Give me a break! LOL! :)

Hi Alegria, I just PM'd you.

Click Here Click Here Click Here Click Here

tell em mike sent ya.

;-)

Bub

SteveStrummerUK

Alegria

Thanks Bud, I'm cool with it now. 

Well, I don't know if I am

I saw the smiley Bub, so I'm guessing your comments were tongue in cheek.

If not, I was actually being serious about the plug-in.

Please don't start turning into a John T 'Mr Censorship' on me.

Hi Steve. :) It was in response to someone's post prior to yours. It actually had nothing to do with your post. And for the record, that software looks great. I may pick it up myself.

The day I turn in to a 'J/J-T' type, you have my permission to put me out of my misery. :)

 
Thanks Bub.
 
All is cool.
 
 

If you do... send me PM.

;-)

Mike,

yea, I use 32 bit OS and the 64 bit engine, same as you.  It gets her done, don't it?

I "got" the essence of Drew's math, although don't ask me to put it on a blackboard.  I just was trying to put it into terms for us mathmatically-challenged, in which category I put myself.  But these threads do seem to go around and around.  The end game always seems come out to if it sounds good, scratch it.

take care,

@

To put the math in perspective, the peak errors I measured mixing 32 stereo tracks using Sonar's 32bit engine were at least 10 to 100 times better than what would be considered "pristine" in the analog world.