true its just the way posts develope, going back to the tangetial thing, your always gonna get a 1000 and one explantions on ****, i think its a good thing as well, if you know wat i mean John i think the confusion over architecture/processing is very common, i know i dont need to tell you that, but its always gonna crop up, that bus is never late lol

its the bit depth as opposed to bit rate Ben, and yeah never mind the bollocks get your mix sorted out and let sonar do the maths lol but aye mate thats why there's no clipping, the maths are applying postive values above the zero, and with the higher bit depth, theres plenty of attention to detail with all the data !!! bit depth = resolution, bit rate = speed of data transfere

Wave


It all depends on what you are looking for in a mix.  But to me it makes a difference.  It's all about the mix, that big sound right? 
 
Anyhow I  see your point but for me I'm very anal when it comes to my mix.
 
Its all up to each individual what sound they want and everyone has thier own tastes.  I'm sitting here chatting when I should be working.  But this is fun   and I don't get out much anymore, I'm getting old I guess.

Ah but this is why I was talking about getting the gain structure right, more head room equals bigger mix, trust me.  People get gain structure wrong and that is why digital can be so crap. IMHO
 
Neb

Ben
 
Right on.

On another (perhaps rambling and only half-thought-out) tangent: the problem that colossal bit depth solves - and 64 bit is truly, massively, making-damn-sure colossal - is one that arises from mixing in a way that pretty much *guarantees* lack of "bigness". It prevents signal degradation over many, many generations of processing. This is is in many ways a remarkable modern development. We can bounce and bounce again, and go through a chain of 20 plug-ins and then bounce some more and compress and expand and bounce again to our heart's content. And we definitely won't get a signal degradation problem per se. What we *will* get is mixes that are drowning in effect soup. If you've ever got a mix that's sounding kind of small and weak, a good exercise can be to see what you can take off. And a good place to start is big complicated effect stacks.

yeah ive read this before, their saying that theres no audible difference and all the bits and high samples rates make no difference, but they are shipping apps that are 32 fp, so why bother ?? lol

John T
 
I do not bounce unless I have to.  You do take a hit on quality every time and it can destroy a mix IF ITS NOT DONE PROPERLY. 

thats my sentiments exactly, let it worry about the maths, and for that reason alone im convincing my self thats its gonna be top notch compared to lesser bit depths, but tbh honest i done a few tracks initally at 24/32/64 and i couldnt tell the difference at 16 what was initially processed at those bit rates, but like i said, cuz of the maths being done, im cool with 64 fp mate

i only ever bounce when im sampling a sound, like programing a kick drum, then load it into reason, i never really freeze tracks, unless the circumstances dictate it

John T


In a vague abstract sense, the 64 bit mix engine is a good thing in that you can just completely stop worrying about this stuff from now until the end of eternity. So on that basis, it's a good and welcome development. But it's not going to bring about a night and day difference in the way music sounds.

You already don't have to worry about it with 32 bit single precision mixing. 

The only reason anyone would worry about it is if they:

1. Don't understand the math.

2. Have been told about vague mathematical errors accumulating during processing.

3. They have never done any objective, quantitative measurements of the magnitude of the errors.

4. They have never done any double blind or ABX testing, so they are easily convinced they can hear stuff just because they know it's there.

5. Because of #1-4, they don't understand that under 32 bit single precision floating point, the errors never accumulate to the point of being audible in the real world. 

6. Someone here in the forum will likely argue with me about this stuff. Almost certainly it will be someone like I described in #1-5.  

John T


Throwing in some tangential food for thought here: it's easy to worry about this stuff WAY too much. There's a lot to be said for just assuming the people building the tools have done their job right, getting on with mixing, and seeing where you end up. If you're getting unsatisfactory mixes, then tracking down and eliminating all your plugs that are processing at 32 bit rather than 64, or whatever, will not make the difference you're looking for. Not saying it's not interesting or worth discussing, but the importance of this stuff tends to get way overstated on forums like this.

Agreed.

And another reason not to worry about it is that any reasonably competent plugin programmer would never use 32 bit processing in situations where 64 bits double precision was more appropriate.

Robomusic


Interesting read ....  http://www.image-line.com/support/FLHelp/html/app_audio.htm

Just skimmed through it, but it does look like something people might want to read.

"The dirty little secrete is PT uses, or at least the last I heard, a 48 bit integer audio engine and to my knowledge no one has complained about having a no. 1 hit with it. ",,,,,,,,,,,,,,,,,,,,,,,,, i came across this white paper a while back, thought it was interesting to say the least !!!!! http://www.sonicstudio.co.../48FixedVs32Floats.pdf

Chregg


"The dirty little secrete is PT uses, or at least the last I heard, a 48 bit integer audio engine and to my knowledge no one has complained about having a no. 1 hit with it. ",,,,,,,,,,,,,,,,,,,,,,,,, i came across this white paper a while back, thought it was interesting to say the least !!!!! http://www.sonicstudio.co.../48FixedVs32Floats.pdf

He's arguing that 48 fixed is a good compromise between 32 bit single precision and 64 bit double precision float (at least in this specific case), though he only actually looks at the performance of the 64 bit version.

Basically, 32 bit float isn't good enough for this scenario (among others) - constructing recursive (IIR), low frequency, high Q filters with +18 dB's of boost and the ability to sweep the controls rapidly without artifacts. 

TC electronics Powercore and ProTools HD both used fixed point processors.

As far as the nyquist thing, I'm not familiar with that program, but I think you might want to start with something like that or a "synth/FX builder" program if you want to concentrate on the DSP stuff rather than the VST part.

I've been playing around with making filters in NI Reaktor, and the filters end up looking like just like the block diagrams shown in that paper. Maybe you could similarly play around in the SynthEdit demo (I don't use it so I'm not sure)? It depends on what your goal is. There's a lot of VST developers on KVR, so you can talk to people there.

And if you want to play around with DSP, you should definitely download Gnu Octave, which is sort of an open source MatLab equivalent (unless you're lucky enough to already have access to MatLab):
http://www.gnu.org/software/octave/

And there are additional packages for it here (especially signal!):
http://octave.sourceforge.net/index.html