So I should also turn my 64bit mix/processing thingy off?

yorolpal

So I should also turn my 64bit mix/processing thingy off?

What I would suggest is do some A/B testing. Don't listen to anything being said here unless you have an identical rig to Danny who has already gone to great lengths to do the tests. If you're doing voice-overs for commercials ... does it really matter anyway? I've never said to the passenger in my car, "Man, did you hear the 64bit processing on that radio commercial?" LOL! :)

It seems to me, if you are using a lower sample and bit depth, for example 44.1/16, the 64bit Engine could be a benefit, but if you're running 48/24 or higher, in my experience, the extra CPU it uses isn't worth the minuscule to non-existent benefits.

yorolpal


So I should also turn my 64bit mix/processing thingy off?

If you're mixing 20 tracks without much processing, you'll almost certainly never ever hear any difference.  

But if you're mixing 100 tracks or something like that (muted ones don't count), with lots of processing on each track, I would probably leave it on, as that's where any of the 32bit single precision math issues that 64bit double precision is supposed to address might accumulate. 

But in reality, with that many tracks and that much processing you're likely to have more audible issues due to other things regarding the tracks themselves, so...

AT

Bub, I have done testing.  A lot of testing over the years.  At home I use 44.1/24 bits.  I have the 64 bit engine engaged.  I find both work well for my purposes, and most pro studios I work at use the same (sans 64 bit engine of SONAR) for rock and roll.

Of course, one should use one's ear and establish what one finds that works best for you.  that's the point.  But people's ears work differently.  I dont' hear distortion using the 64 bit engine.  You do.

I never said I hear distortion using the 64bit Engine. I actually said the opposite in that I did a null test and there was no difference between a project exported with the 64bit Engine on or off. Why did you say that AT? :( It seems like your whole post is based off that misunderstanding. :(

Here's the problem with threads like this ... everybody thinks they are right because they go around dropping names, and spouting theory and when someone goes against that, they get offended. Theory is just that ... theory ... and honestly it really doesn't matter because you only have so many options to choose from anyway. Theory aside, there are only so many M-Audio's, MOTU, RME's, in the world.

What I'm saying is, do some null test with and without the 64bit Engine on. Do some with different types of dithering. Do some with different sample rates, which is how I found out that my M-audio Fast Track Ultra, not a MOTU like you misquoted me on, sounded better at 44.1kHz.

You'll be shocked at what you find ... which is very little difference between any of it.

Maybe my ears or too old and abused (I listened or mixed a lot of R&R in my younger days and even in my early 20s I was losing high-frequency in my left ear).  Maybe the distortion I don't really hear I like.  Who knows.  But I don't find it hurts anything.

The science behind sound and systems isn't an answer in itself, but it can't hurt, either.  And no, I don't pay much attention to tom dick and hairy on any forum.  But if Dan lavry sez that a sample rate 60-70 kHz is optimum and makes a cogent arguement why, I pay attention.  It is nice to know some of the rationale behind decisions others make, scienctific or otherwise, but only to make choices for yourself.  You say your Moto sounds better at 44.1 than at higher rates, which doesn't make any sense technically.

Most designers say the difference (esp. in older gear) in sound between lower and higher frequencies stems from the fact that anti-aliasing filters can have a lower slope at higher rates, which makes perfect sense.  You hear it differently, and no one can tell you what you like.  But I notice that you use the higher rates for latency reasons, so the sound itself must not be that different. 

That is Danny's point, I think.  Most of these bit and sample rate choices don't make that much difference - only around the edges.  It won't make a bad song good.  It is like expensive gear.  Engineers, like wine tasters, are after that elusive last percentage point that the average drinker or the listener won't ever know, and what's more, don't care about.  But for engineers it is their business or passion so they spend the bucks and muck around to get what they hear as "best."  And even muck around on forums like these to pick up some tidbits.

@

Let me know how you like the VRM...I find it a useful tool.

x64bit audio Engine!


It is a different in sound and I can clearly hear it in my STUDIO. If you have trained ears and use of high end Monitors and AD/DA converters you should hear it too. 
IF its good or Bad is another story..?

What it do is INTERNAL MIX-ENGINE, ditherring EQ and plugins internal to 64bit VS 32bit.. 
I can be good but can also be very bad. Example if you use highend filters and EQ plugins and highend soundcards you don't want anything INTERNAL color you filters or sounds and plugins...with dithering.. 
If you like it use it.. if you don't, don't use it,,,

Often Dithering comes later during MASTERING anyway...

Bub, just walk away and let the "expert" deal with it. I worry about you blood pressure buddy.

shawn@trustmedia.tv


Ok, so, I have a 64bit laptop with and use it's internal soundcard with ASIO4ALL I use for mixing and making VST (mostly) music, live recording is done with NI AudioKontrol 1. What is the most efficient, 44.1 16 or 48 24 or something else that doesn't "real world" lose sound quality. I believe I've heard in these posts that a higher sample rate decreases latency so I'm thinking of switching from everything set to 16 bits 44.1 (good sounding,low CPU/harddrive use is what I thought) but it looks like I'll get lower latency and more headroom at 48 24 without sacrificing my CPU Harddrive much at all (it's a beefy laptop). Someone give me a push over the edge, I think I'm inspired enough to start a new project and change all my settings!

A recording bit depth of 24 is the key to a low noise floor and best headroom.  The OP's question had already been solved by John and Mikes first two responses.

As for the rest a 44.1 kHz sample rate is fine for CD or Digital download media, 48 or 96 if you are recording soundtracks or for high end media where as Danny says your input source is clean enough to warrant it, or as Bub points out to reduce your latency (if your machine is up to it).

All the other settings for internal processing and rendering you can happily set to suit your own preferred set of particular beliefs and any ignorance of what is truly going on here isn't going to cost dearly.  Bub again has clearly highlighted that.

I use 24/48,000 simply because my old E-Mu card behaved better at that sample rate and it's become a bit of a habit for me now, other than that I'd be quite happy at 44,100.

Good posts from both Danny and Bub I reckon.

AT


That is Danny's point, I think.  Most of these bit and sample rate choices don't make that much difference - only around the edges.  It won't make a bad song good.  It is like expensive gear.  Engineers, like wine tasters, are after that elusive last percentage point that the average drinker or the listener won't ever know, and what's more, don't care about.  But for engineers it is their business or passion so they spend the bucks and muck around to get what they hear as "best."  And even muck around on forums like these to pick up some tidbits.

@

That's exactly what I was getting at, AT. I have found in my years that people seem to buy into the hype of some of this stuff and *think* that can really hear a difference. When you do the right testing to where you are not looking at what you are listening to, it changes the entire ballgame...it truly does. I can't tell you how many times I thought I could tell differences. Yet when I did blind tests over and over again and failed, I found out the appearance of a name had all the power moreso than what my ears thought they heard.
 
Also keep in mind that the purpose of my post really wasn't do discredit anyone or make anyone feel bad. I've only explained what I have lived the same as I do when someone comes out with some recording technique they think is the master of all decisions. I try it over and over again to see/hear if it is beneficial to me. If it doesn't make me go "wow" it's just another tool for the box. If I didn't hear a difference between 16/44 and 24/48, rest assured 16/44 is where I'd be. Weird thing with that is....I actually recorded a project for a client that needed guitar tracks. I used one of my net boxes which has a Realtek in it that records at 16/44. These were meant to be scratch guitar tracks just to see if the client would like them. They sounded so good (and actually warmer than 24/48 go figure?!) that we kept them. I was so upset, I decided to redo the guitars on the good stuff only to find out for this particular speaker sim tone, the 16/44 Realtek was actually the better choice.
 
So I sincerely think it's important to use what is best for each person really. But to make the right decision, it's good to test things out blindly so that you truly don't know what you're listening to until it's time to find out. For example, my experiences with null testing have yielded strange results. Sometimes the one that is incorrect actually sounds better to me. It depends on the song and the sound you're looking for. So for me, a test like that wouldn't suit my needs....though I can see how it would be incredibly helpful in other ways and situations.
 
I just think like this....24/48 is a really good set-up. It's digital...you're not losing anything if you bounce or export. 32 bit float and the 64 thing supposedly make sure you don't lose anything. LOL! That just doesn't make sense to me really. Though some of the math gets jumbled in what we do everyday, I have never heard something so off and wrong that I had to redo it because "my math was off". Seriously...think about that for a second. How much truncation would there need to be before we decided "ok, this mix actually sounds bad all of a sudden"? In all my years, everything I record digitally is what it sounds like. If I get the sound right pre-print, it sounds good during mixing and after it's been exported. If my card can't go higher than 24 bit, I'm kinda adding something that doesn't exist, right? I mean seriously...that's how I view it whether I'm right or wrong. The 64 bit engine thing....I just don't get it. I know what it claims to do....I just am not a believer in adding anything that really doesn't make a huge difference.
 
It's like buying an expensive mic pre and doing a shoot-out with a cheaper one. Sure, some people may hear a difference...some may not...is that difference worth $5000 extra dollars? Are the bit and samples as well as floating points worth the extra processing and resources? To some, they are and I'll never dispute that or call anyone on it. For me, it's absolutely useless and a matter of hype and some dude's math problem not coming out right that makes no difference on how good or bad a song is or whether of not the wuality has degraded. None of that changes THAT stuff and to me, that is the most important. :)
 
-Danny 

In speaking about 16 bits I think its under rated. Well done 16 bit recordings are for my money just as good as 24 bit recordings. But for me 24 bits are so forgiving that its not worth the trouble to get a great 16 bit recording when 24 bits does it with ease. Sample rate is another matter. High sample rates only give wider bandwidth nothing more. There is no greater precision with them over 44.1 or 48 for the bandwidth intended. This point has been proven over and over again. Yes some interfaces will perform at a given sample rate better but that has nothing to do with the viability of a lower sample rates for high quality recordings.

Bub relax, this disscusion was over ages ago, yes the anal sex thing is a little crude but let it go.  If anyone had read my OP near the begining of this topic, I gave the theory of why to record at 32 bit floating point and there is at least one interface that allows you to set the driver to 32 bitFP that is Lexicons US42s (yes I have owned one).

I also said in my OP that this theory was yet to be proved, I will take your experiment at face value but I think a lot more investigation needs to be done before we can put this one to bed.

No amount of bickering between hobbiest, pros and all the rest that inhabit this forum is going to clear this up.  For me I can feel, see and know there is a difference when recording at 32FP, IMHO, so I will continue this practice until proven otherwise.  This is a not a slight at you bub, I have looked over your experiment but we need the audio scientets to set up a real world experiment to nix this or to proove it.  I am not an audio scientest, but I am aufauit with the basics.

Peace Ben 

"aufauit" with the basics??  Is that Launceston jargon?

I never said my French was any good Bonjour quest see that is it!!!!  I tried to study french at uni, didn't even need to havd any work in for the first semester and I still failed!!!

Peace

Who's doing these math problems?

yorolpal


"aufauit" with the basics??  Is that Launceston jargon?

Jargon in Launceston?

Nah more likely a Corny Pastiche.

Not unlike this one...