AT: totally with you there. The stuff is definitely good to know and there are certain situations where something you may not think will work right, actually does. LOL! I can't tell you how upset I was that a Realtek actually gave me better results than my good stuff. But, for some things...I guess it's acceptable. I do notice that the layla or the rme kill on anything that is mic'd. But for the direct guitar stuff or speaker sim stuff...I just may use that Realtek more often for dirty guitars. LOL!
 
John: Well said...I have proof of this also. I get so many projects in here at different bit and sample rates for clients, that sometimes AFTER I'm done with their stuff, I forget to put my drivers back to 24/48 for myself. I've left things at 16/44, 24/44, 24/88 and 24/96 and recorded without thinking. Some of the best recordings I've done were actually 16/44. However, there is definitely a difference for me at 24/48. But....when you set up the right sound sources, 16/44 is going to do a fantastic job. :)
 
-Danny

John

John: Well said...I have proof of this also. I get so many projects in here at different bit and sample rates for clients, that sometimes AFTER I'm done with their stuff, I forget to put my drivers back to 24/48 for myself. I've left things at 16/44, 24/44, 24/88 and 24/96 and recorded without thinking. Some of the best recordings I've done were actually 16/44. However, there is definitely a difference for me at 24/48. But....when you set up the right sound sources, 16/44 is going to do a fantastic job. :) -Danny

That sounds like a royal pain in the rear. 

Totally agree with your points.

I do think proper micing and attention to detail will go far further then what the bit depth and sample rate are.

LOL yeah sometimes it is a pain. But I draw up cue sheets for each client as well as "work to be done" instructions for myself. This way I never forget anything for THEM. But I seem to forget things for ME. Hahaha! It's ok though...I'd rather forget me than mess something up for someone else. When I ask some of my clients why they record where they do...the answers are something else, John. Here's a few I've gotten over the years. Keep in mind, I'm not making fun of anyone...it just proves how even some incredibly great engineers respond...some of which, I'd apply for a job as their water boy because they smoke me. I did of course respond with "you know you really don't have to do that" and "just try this for me". :) Some listened and tried, others stuck to what they believed in and it's fine either way really. But check some of these out, and honest I'm not making these up.
 
"That's where it was set when I bought it"
 
"I figured the higher I can get this baby to go, the better it would sound"
 
"C'mon Danny, what type of studio you running there...everyone runs at 24/192 you rat baystid"
 
"Ah, I get asked that question all the time...I can just feel 24/88 in my bones more...like it hits you in the heart"
 
"24/44 is the only setting I could get to work right in my DAW"
 
"24/96 just sounds more crisp"
 
"16/44 is what CD's are....why should I go higher and dither all the quality away? By the way, what IS dither?" (that one actually made me chuckle and took some explaining on my part)
 
Some of these cats get sounds that would make your head spin. It's amazing how much they don't know about as far as aesthetics go and how they are trial and error guys that just know how to record and mix great albums...but guess what, I'd put myself in that same category...I don't know much science or theory and though I can't compete with some of these monsters I'm talking about, I do a pretty decent job just from doing it for so long using my ears and life experience. Whatever works is fine by me...as long as you get your ideas out of your head and have a great representation of what you're about, it's a great thing....even if you record at 8/22. LOL! :)
 
-Danny

Benefits of 64-bit Mixing engine 

http://www.youtube.com/watch?v=-9EeW9WhNWA



Here is VIDEO that answer and prove the QUESTIONS some of you have...

Now I want to shoot myself, or perhaps for the sake of double precision accuracy get somebody else to shoot me.

Anyone have the formula for most efficient propellant compound to get the sure fire amount of lead into my head to get the job done?

Will it even matter?  I'm reckoning 0.50 calibre will cover any margin for error.

mike_mccue


Would you settle for an oven?

No, I want to hear a 'splat'.

http://www.trustmedia.tv/House_of_the_Rebirth.mp3

All this talk about bit depth has made me achieve lower latency...My DAW smokes...

New song in Deeper Depth & Higher Res:

http://www.trustmedia.tv/House_of_the_Rebirth.mp3

 
 

shawn@trustmedia.tv

New song in Deeper Depth & Higher Res:

http://www.trustmedia.tv/House_of_the_Rebirth.mp3

 
 

BANG!<---0.6ms-->(splat!)

I never switched my dither, still none...I will give it a try!

Freddie H


Benefits of 64-bit Mixing engine 

http://www.youtube.com/watch?v=-9EeW9WhNWA



Here is VIDEO that answer and prove the QUESTIONS some of you have...

Thanks for sharing that Freddie, but wouldn't you admit that...that sure is a lot of math that would literally do nothing to alter the sound of a project drastically? I've used Sonar for many years at 24/48 with my Layla cards and 32 bit float and 64 bit mix engine did not exist. How on earth could someone literally use that video as proof of anything but a math equasion that might be wrong? We're audio engineers....we've used tape and made great sounding records for years...some of which will blow the stuff done today out of the water. See, this is the stuff that burns me up when I get involved in discussions like this...this is the hype I'm talking about that will make a 2% difference MAYBE or less. If we could measure analog tape to digital using one of those math equasions, just how messed up would the math be for what we may consider "one of the best albums ever produced"?
 
I dunno, something Steely Dan, Doobie Brothers, insert your fave vinyl album here. How bad would the math be off? I'd be willing to bet horrendously off, yes? If it could be measured (which maybe this stuff can, I have no clue about any of it) what would the actual bit/sample rate of an album be? Anyone know? If the best albums we could all come up with were somehow measured, what do you think we're listening to...12 bit 20? LOL! They sure do sound great whatever they come out to....bad math or not.
 
Personally, I just think judging whether to use something or not based on a math problem is ludicrous in the audio field unless it truly gives you a major loss in fidelity that is heard by all or at least a majority of people. Why on earth would we want to judge noise as our test subject? How great can you eq noise to sound? Granted, recording audio IS noise in theory, but it's quite a bit more pleasing to listen to in its worst form than pink noise or white noise or whatever other noise you want to test with, and a bad recording can most times be salvaged to sound good. Noise, will always sound like noise. Who cares if the math is wrong on that? Seriously....I just can't believe some of the stuff I read. LOL!
 
So you're telling me that if I post up 2 mixes on here...one using whatever I want, the other using 32 bit float and 64 bit mix engine...it's going to be a huge difference that everyone will hear that will make the entire forum switch over to it? See, I think it's much simpler than that and all this talk is complicating the obvious. It's cut and dry really. You get a sound source to sound good....record it digitally and what you get is what you got. Eq it, compress it, effect it, export it...done. You either know how to do that correctly or you don't.
 
How many bits or samples could we lose to where it totally affects the audio and makes it sound like you don't know what you are doing? I'd bet not enough to make a huge difference.....because I sure can't hear it...and if *I* can't hear it, I'd bet all that I own that most common listeners won't be able to hear it either. Who cares what other engineers think....I don't produce projects for other engineers unless I'm asked to specifically. I produce projects for music lovers.
 
It's like me creating what I feel is the greatest guitar sound known to man. No one but me, guitar players and a few music lovers of guitar will appreciate it. The common folks will not care....they won't care if I tap like Eddie Van Halen, sweep pick like Paul Gilbert or run legato like Alan Holdsworth. None of it matters to anyone but those buying into the hype.
 
-Danny

mike_mccue


It's easier to just record everything at 16bit.

Then you don't have to waste all that time listening to all the dither renders trying to figure out which one is better.

Yes, it saves a lot of dithering.

mike_mccue


Danni,

What is your point?


I thought most of us just used the 64bit mix engine from the moment it became available nearly a decade ago.

I guess not?

I don't know of any downside to using it. (well maybe a few bugs along the way with some overclocked systems... but that seems all worked out.) I turned it on the day I got it and haven't thought about it since. No problems.

It is demonstrably better in terms of rounding errors. That is a very simple scientific fact.

So...       So what if you may not be able to hear a difference? It doesn't seem to serve any purpose to draw a distinction about something that is as simple as checking a check box and moving on. It doesn't cost any extra money... it's been there for years... it works... it sounds good... there is sceince behind it... it doesn't cost any money... it works... what is the big deal?

I don't think there is any way to comment on the analogy to Steely Dan and the Doobies without breaking the *discussion* down into a basics of digital audio review... and I don't want to go there. :-)

all the best,
mike

Hmm, I thought I explained my point quite well, Myke? I used the 64 bit mix engine when I got Sonar 5. It didn't make a difference I could hear that made me go "wow". It also spiked up a few cpu cycles on my pc at the time and gave me a few clicking noises. I did eventually make it work using Sonar 7 on another machine. Am I so wrong to think it didn't make enough of a difference to use the extra resources and share my experiences with people? 
 
Explain rounding errors and tell me how they will make my audio sound horrible compared to not using this stuff please? Rounding errors to me...(being someone that knows recording probably near as well as you know about your camara work).....inability to record properly. Show me proof of this scientific fact degrading the quality of audio where I can hear it for myself while an advanced engineer is at the helm.
 
Why check a box that doesn't make a difference to me just because the math is wrong? Please prove to me that there is a legit audible degradation on your end and I will prove you wrong by doing a project using all the wrong stuff and make it sound presentable. It works to you and a few others...that's great, your ears are better than mine and I'm jealous....what else can I say?
 
Big deal? I spoke my mind on how I've lived and experienced it. There is no big deal other than you having a problem with my opinion from years of doing this vs your years of doing this.
 
How would the discussion break when I merely asked simple questions? You don't have to break anything down or change the discussion....I asked what bit and sample rate those albums would be considered if they were able to be measured. It doesn't take science, if you know the answer, why not just give it to me? Is it less than 16/44? I would believe it would be. If so, and you could justify any type of math with it, how bad would the numbers look? Bad I would bet....so, if what I'm assuming may be the case and I do not have a clue for sure, to me that is proof right there that incredible albums existed with errors and lesser bit and sampling quality that sounded incredible.
 
With what we have today...which is way more pristine and unforgivable, how on earth could these wrong math problems totally degrade our audio? This is what I do not understand. I will not buy into the fact that using 32 bit float and the 64 bit mix engine will make my songs sound better to the point of going "wow". I will not buy into the point that my incorrect math within the files of my music will make my songs sound bad. Why is that so hard to understand and why would you even call me on it?
 
So, why not show me audible proof? Since I can't hear it on my end when I've tried it, show me how this bad math is messing something up and please forget about the science of it and give me something my ears can listen to. If you can prove the point, and make it legit to where I can duplicate it, I'll buy into it and give you props for teaching me and will use it exclusively. But until that time, I have to chalk it up as hype and nothing that will make THAT great of a difference.
 
Ok, I'll make it simpler.
 
1. Does our audio degrade to where it can be heard by the human ear from not using these little check boxes and having bad math?
 
2. Will our audio improve by using these check boxes, and if so, will it be audible enough for the human ear to hear? How much of an improvement....2%, 5%, 10%?
 
3. If your answer is yes and yes, could you please supply audible proof of this as well as showing me the steps you took in order to degrade your audio?
 
-Danny

Slither your dither?

I'm with you Mike.  I turned the 64 bit engine on and haven't really thought about it sense.  I would assume that it would use more cpu, but never had a problem w/ my elderly system using it.  I don't go in for the 100 tracks behemoths, either, and freeze my softsynths as I go along.  I never heard a problem and I don't think it is hurting anything, so why not?

But again, if it caused a problem to click etc., I'd be wary of turning it on and keep a big stick handy.  I mean, there are plenty of other things to worry about when recording than if my system is going to crap out.  Things like is this the right water for the diva ... I mean singer?  Or is the guitarist going to play too many notes again.  Finaly, is someone also drinking out of my beer?  The important stuff, you know.

@

Freddie H


Benefits of 64-bit Mixing engine 

http://www.youtube.com/watch?v=-9EeW9WhNWA



Here is VIDEO that answer and prove the QUESTIONS some of you have...

Unfortunately Freddie, though what he says is technically correct, he significantly cheats at the end.

He calculates the 3dB to 6dB error you can get with a minimal number of tracks, says something like "this 1 or 2 bit error may not matter", which is 100% correct. But then he says something like "I'm not a golden ear guy, but people who turned it on say it sounds great!".

IOW, the math in his particular test only shows a 6dB maximum error, but it's reported that subjective types think it sounds great.


In reality, if the error is only 6dB in 24bit audio, that equates to -138 dBFS, which is never, ever, ever, ever, ever, ever, ever, ever, ever, ever, ever, ever, ever going to be audible.  

It's fairly easy to create a test that simulates the error accumulation from parallel or serial operations for various numbers of operations.


The question that needs to be asked is:

What are the lowest peak and average errors any of you think are audible relative to full scale (in either real world or ideal listening conditions)? 

If you do the calculations, you'll find that you need quite a few tracks/operations for even the peak errors to accumulate to even something like -120 dBFS. And note that for every 6dB below 0 dBFS a given sample is, the error level is reduced by 6dB for that sample.

Also, Drew... can you relate your previous post to a 16bit final output scenario?

Do the errors just get truncated in the end?

best regards,
mike

mike_mccue


Also, Drew... can you relate your previous post to a 16bit final output scenario?

Do the errors just get truncated in the end?

best regards,
mike

Yes.

And keep in mind you'd need to be unlucky enough for your peak calculation errors to both be ~+48dB above silence in floating point and happen to fall on a particular sample >= ~-6dBFS for any errors to make it into 16bit audio.