my previous build of the 32-bit xp machine had the soundblaster live drivers built into the os install. since putting the live back in the 98 machine (which i'm calling 16-bit but is actually also 32-bit) where it came from, i've put the audiigy that my sister's first husband gave me (he decided to be a douche bag mac user around '05ish, so he didn't want it) in the 32-bit machine to replace it. it also has what was the last line of m-audio delta series pci cards in it, which i got for very cheap at the end of it's lifecycle explicitly because it has an actual rca out on it, and the dac is dealt with right away kind of thing - it's about minimal amounts of conversion, although i later found out that the windows kmixer sort of ruins it and i have to connect via asio to make sense of that. and, i'm connecting an alesis 16-track mixer as a recording interface via firewire, as well as a pod xt via usb for direct guitar ins. old gear - like i said. and, none of this is going anywhere any time soon....
anyways, i didn't need to script up a driver install for the live because the drivers came with the os and the software didn't work on xp anyways, but i'm going to need to script something up for the audigy. i'm currently searching for an install cd, and will probably hold off on anything complicated until i can get one of those interfaces like i have in the other machine. the major reason i'm going to use this device is to make use of the dsp in it, which is intended for gaming and is consequently a little thicker than you'd get out of most normal audio-focused applications. a large percentage of the weird guitar and drum reverb i used from about 1999-2004 was actually using that eax as a pre-amp; i lost it when i upgraded to xp in 2004 and put it aside when i couldn't get it to work when i built the 32-bit machine in 2007.
wait. i'm recording at 16 bits? what? well...
see, this is my actual logic, here - the machine i did everything on until my system collapse was built to use 32-bit software, and almost all of the plugin software (like guitar rig) is written for 16-bit audio manipulation. so, what happens when you send 24 bit audio into software written for 16-bit audio? if it's well written software, it should actually stop you from actually doing it and tell you to downsample the audio or buy a new plugin. but, what most plugins are going to actually do is simply truncate the file and then spit it out with a bunch of zeros, meaning you end up downsampling your 24 bit recording to 16 bit by accident as you process it. and, you probably will not be able to tell the difference, as you're doing it.
on top of that, you downsample in the end anyways, right? see, that's the assumption. is it true? for now, let's assume it is. now, some people will tell you that you want the extra space for the calculations, if your software allows for it. ok. but, that's the rare scenario where you probably can tell the difference between 16 and 24 bit - when you've run the project through a thick reverb and filled all that space up with data. and, then what happens when you mix it down?
the answer is that you just delete it - and you've wasted your time getting something to sound great as a 24-bit sound file, only to have it sound entirely different as a 16-bit product. and, are you going to even notice, are you going to listen to a 16-bit file through your shmancy sound system? nooooo.
oops.
so, i developed a policy of sticking to recording to the output quality. that is, i decided that if people are going to listen to the end result on 16-bit playback devices (or, god forbid, fucking speakerphones) then i'm going to mix them at 16-bits. logical, right?
but, how true is that, still? i mean, nowadays nobody listens to anything on cds - what they actually do is stream it, probably. and, are they therefore able to stream it at 24-bits? i mean, their hardware can probably do it.
the 32-bit machine that i'm reinstalling this morning will remain 32 bits and geared towards 16 bit recording. that's not going to change.
but, as i get around to building that 64-bit machine and wondering how i'll be recording into it, the question i'm actually going to ask isn't about maximizing recording quality so much as it's about trying to understand actual listening habits. i know that i still listen to everything in 16-bit. am i out of touch on this? if i am, and people are listening to 24-bit flacs on their phones, then i should adjust to mixing final outputs in a way that conforms to how they're actually going to listen to them.
if i mix everything in 24-bit and the world listens to it in downsampled 16-bit, i'm just deluding myself through the process - all of that effort exists strictly in a parallel reality that nobody actually inhabits. however, if i mix everything in 16-bit and the world upsamples it to 24-bit to listen to it, i'm missing out on an opportunity to expand the quality of the mix.
all the technicalities and specs are fun to analyze, but it's the user experience that i'm really concerned about.