I record, mix and master in 96khz then export to 16bit 44khz for CD or 48khz for video.

I'm the same as Marco, with the exception that for most pro bono work I will run at 44.1KHz throughout.

I do 24/48 bc a very talented person that I trust, told me to lol. I think with what I've read on here, I'll probably start doing (with new projects) 24/44.1

Brother Junk, post: 446561, member: 49944 wrote: I do 24/48 bc a very talented person that I trust, told me to lol. I think with what I've read on here, I'll probably start doing (with new projects) 24/44.1

Make some tests for yourself. Results vary depending on the converters, the DAW and plugins you use.

Get this, even if you don't hear the difference, the mastering engineer might (if you send your work to one)
Also, when you tell a customer you record at 24/96, they at least have the impression you want to capture the maximum details of their music (psychology works) ;)

But frankly, I tried it for myself and with the tool I use, 96 sounds better to me...

pcrecord, post: 446564, member: 46460 wrote: Get this, even if you don't hear the difference, the mastering engineer might (if you send your work to one)
Also, when you tell a customer you record at 24/96, they at least have the impression you want to capture the maximum details of their music (psychology works) ;)

But frankly, I tried it for myself and with the tool I use, 96 sounds better to me...

I would use 96 if it would allow me more than 3 tracks and 3 plug-ins. Imo, it sounds better. But for whatever reason, my setup doesn't want to do 24/96.

I'm curious as to what the big boys are doing in those 3 million dollar rooms... are they also at 44, 48 or 96?

Or are they typically at the ultra-high SR's, like 192?

One would think that those rooms have the ultimate in computer processing horsepower and storage.

Just curious...

DonnyThompson, post: 446587, member: 46114 wrote: I'm curious as to what the big boys are doing in those 3 million dollar rooms... are they also at 44, 48 or 96?

Or are they typically at the ultra-high SR's, like 192?

One would think that those rooms have the ultimate in computer processing horsepower and storage.

Just curious...

I would also be curious...

DonnyThompson, post: 446587, member: 46114 wrote: I'm curious as to what the big boys are doing in those 3 million dollar rooms... are they also at 44, 48 or 96?

Or are they typically at the ultra-high SR's, like 192?

One would think that those rooms have the ultimate in computer processing horsepower and storage.

Just curious...

I bet they can afford to run at 48 or 44 because quality of their converters, but I'm also curious ;)

I might be wrong but I'm on the impression you get a wider quality difference between 44 and 96 when using a budget interface.
But on a highend converter the difference between 44 and 96 is smaller.
Still, a highend at 44 sounds better than budget one at 96...

pcrecord, post: 446589, member: 46460 wrote: I might be wrong but I'm on the impression you get a wider quality difference between 44 and 96 when using a budget interface.

I've only seen a couple examples of this, but my guess is you are right. My Mbox Pro, it makes a difference. I can't actually use 96khz because my track count drops to almost nothing, but it definitely sounds better than 48. But the studios I know often use 24/44.1. But they have much better converters.

Previously, I would stay 44.1k throughout, and only go to 48k if it was a project strictly for video standards. It's only within the last year or two I've had the capability to track at 88.2kHz or 96kHz, so I've had some time to think - which clearly can be dangerous.

I've been doing my own version of the 2 DAW mix down process since the mid 90's, but perhaps, Boswell could clear something up for me regarding rendering a mix down to stereo ITB, when necessary.

If I know the final product is destined to be a CD at 44.1kHz, is there any advantage to recording at 88.2kHz, rather than 96kHz?
The change in bit-rate is unavoidable going to CD, but does tracking at 88.2kHz minimize the potential for errors in the sample-rate conversion algorithm since it's divisible by a nice even 2.0, rather than 2.17687074829932?
It seems as though the error-correction would have to truncate a lot of data, all the while 'dumbing down' the bit-rate as well. To my way of thinking, that seems like a lot of inexact math that would create cumulative error over the timeline. Whether it's audible or not, would be a separate issue.

In the more tangible realm, I can see a difference in definition (not just image size) when reducing a hi-res digital photo by 50% or 25% vs. 66.666% or 33.333% in Photoshop. It's one thing when you tell 4 neighboring pixels to pick an average color and combine into a single pixel. It's another to tell 5 ⅓ pixels to average and combine - with so many remainders and compromises. Does any of that carry over to digital audio, where we're essentially telling 96,000 samples to merge and occupy the same amount of time as 44,100 samples? Does that happen more accurately with an even divisor, or is it irrelevant?

Am I missing something? Am I wrong on the fundamentals? Am I too dumb to know, how much I don't know about it? <- causing me to oversimplify it? Is the error-correction (if error even exists) so good no one would ever hear it?

(I will gladly concede any, or all of those points.)

Thank you sir in advance for generously educating knuckleheads like me with the curiosity of how things work, but not nearly enough of the schooling!

dvdhawk, post: 446602, member: 36047 wrote: If I know the final product is destined to be a CD at 44.1kHz, is there any advantage to recording at 88.2kHz, rather than 96kHz?
The change in bit-rate is unavoidable going to CD, but does tracking at 88.2kHz minimize the potential for errors in the sample-rate conversion algorithm since it's divisible by a nice even 2.0, rather than 2.17687074829932?

Many were babling about this for years and I was one of those to record in 88.2 for a while.
In my recent tests, I honestly didn't hear the difference between going to 44 from 88 or 96 so I decided to keep 96 for all my projects video and audio.

dvdhawk, post: 446602, member: 36047 wrote: Does any of that carry over to digital audio, where we're essentially telling 96,000 samples to merge and occupy the same amount of time as 44,100 samples? Does that happen more accurately with an even divisor, or is it irrelevant?

Now I don't think it's a merge really.. I'm guessing there is some sample dropping and averaging but merging ?? Frankly I don't know enough about that subject. :cautious:

dvdhawk wrote: Previously, I would stay 44.1k throughout, and only go to 48k if it was a project strictly for video standards. It's only within the last year or two I've had the capability to track at 88.2kHz or 96kHz, so I've had some time to think - which clearly can be dangerous.

I've been doing my own version of the 2 DAW mix down process since the mid 90's, but perhaps, Boswell could clear something up for me regarding rendering a mix down to stereo ITB, when necessary.

If I know the final product is destined to be a CD at 44.1kHz, is there any advantage to recording at 88.2kHz, rather than 96kHz?
The change in bit-rate is unavoidable going to CD, but does tracking at 88.2kHz minimize the potential for errors in the sample-rate conversion algorithm since it's divisible by a nice even 2.0, rather than 2.17687074829932?
It seems as though the error-correction would have to truncate a lot of data, all the while 'dumbing down' the bit-rate as well. To my way of thinking, that seems like a lot of inexact math that would create cumulative error over the timeline. Whether it's audible or not, would be a separate issue.

In the more tangible realm, I can see a difference in definition (not just image size) when reducing a hi-res digital photo by 50% or 25% vs. 66.666% or 33.333% in Photoshop. It's one thing when you tell 4 neighboring pixels to pick an average color and combine into a single pixel. It's another to tell 5 ⅓ pixels to average and combine - with so many remainders and compromises. Does any of that carry over to digital audio, where we're essentially telling 96,000 samples to merge and occupy the same amount of time as 44,100 samples? Does that happen more accurately with an even divisor, or is it irrelevant?

Am I missing something? Am I wrong on the fundamentals? Am I too dumb to know, how much I don't know about it? <- causing me to oversimplify it? Is the error-correction (if error even exists) so good no one would ever hear it?

It's entirely up to the programmers who code up the DAW. All the DAWs I have examined use a generalised algorithm that employs a single technique for SRC that does not take advantage of simple multiples when they occur. There may well be others that spot the special cases and take short cuts.

I did the investigation before I standardised on 96KHz and a two-box mixdown. It helps the two-box mixdown process not to have sampling rates that are nominally exact multiples, as the crucial point is that the source and destination are SR uncoupled.