What bit rate do you record/mix at?

Quantifying musical experiences is difficult - but surely it's a part of our job? Just like when we tell them something "wont work"  like stacking drop tuned bass guitar lines in diatonic thirds. Understanding why these things happen, and when, can really help you predict problems.

Again, no one disagrees that there might be differences within a 96Khz recording. The problem is quantifying that - is it Ultrasonics? Is it Aliasing distortion? Filter ripple? Or is it simply confirmation bias? Have you ABX'd it?

I've heard claims that it's subsonic and sub bass frequencies that really benefit from oversampling as our own hearing system expects to hear very gradual rate of change in such signal amplitudes. A very interesting theory that has a very wide set of implications. However, I've never seen proof and ABX's don't show anything so I can't bank on it. It's not a good use of my own limited resources.

And we do have more science outside of the scopes. A lot of great psychoacoustic research out there, like on Masking and Localization and critical bands of pitch and amplitude perception. I've a whole library of psychoacoustic tests for a listener if people wanted an upload - it's frankly fascinating to hear how your brain reacts to filtering out and reintroducing harmonics, putting noise over tones, detuning tones by minimal amounts, detuning harmonics, linear vs. logarithmic scales etc. etc.

..And some weird crap like this:http://en.wikipedia.org/wiki/Deutsch's_scale_illusion

Yes jazznoise,
Me and my bud John, have blind tested a ton of shit together, not always the most scientific approaches, but i've always found that to be kind of a waste in an intimate and familiar setting, it's either happening or it isn't. We did some blind testing of some tracks, none more obvious than acoustic guitar tracks at 96/24 and 44.1/24 on my bla 002, we write down our answers, and we both blindly picked 96 every time, there is more density and richness in the sound to my ears at the higher sample rate. Believe this, i'd be happy if i could use less dsp and storage space with a 44.1 sampling rate, i'm just not willing to take what i perceive as a sonic hit. The bottom line for me is, i'm trying to create the best sounding system i possible can, and if i can get one iota of sonic improvement, i'll take it every time.

I cannot wait to finish this new rig, i'm starting to see a little flicker that just may be a light, at the end of this long diy tunnel?

cheers
T

I can add a few comments. I see the responsable sample rate to use in terms of the future financial value of a recording as being 48 or 96.

This is because CD players are going away leaving the principal playback media being video and audio files. At least to me converting from 48 to 44 sounds better than converting from 44 to 48.

I try to record at 96 if its practical. Otherwise I use 48. My tests have shown 96k to improve the sound of both 48 and 44.1 final products. I've also found a number of plug-ins sound better at those sample rates than at 88 or 44.1.

44.1k 24bit because I'm not a fidelity, A/B measure meters in millimeters kinda guy.

I've always kind of wondered what the difference 48 from 44.1 really made...If I'm not recording in 88.2, I usually just go 44.1

IMO, If it grooves at 44.1k, it grooves harder at 96k, it ain't millimeters of diff on my rig, it's significantly sweeter to my ears, ymmv, so what ever's clever

I'm in sir Bobs camp on this one

I would ask those who swear by 44.1, have you ever actually recorded at 48khz? 88? 96? More than one track once, didn't hear a difference and moved on?

I'm always interested in the road that brings people to where they are as much as where they are.

For me, the decision is pragmatic. I would work at 88.2 or 96K but my Mini starts to run out of CPU, partly due to having BFD3 instantiated at 128 samples while tracking drums. 48K ends up being the highest stable rate while tracking MIDI on my rig, so 48 it is. I can hear the difference between 44 and 88/96, but not really between 44 and 48. But I'll take the theoretical bump in quality with a little CPU hit any day.

Obviously nobody buys media based on sample rate however all video uses 48 or 96 so 88 always creates a need for additional conversion to work with picture. Today most people's best sounding system for engaged entertainment is video and not audio like it once was. MP3 and AAC both sound better at 48 and AAC sounds incredible at 96.

Gotcha. Licensing to video. I had never considered that being a music director's mind...but...that seems good enough reason. 

You're not likely testing what you think you are with that test, though. But, are you seriously suggesting I couldn't tell an mp3 from a linear redbook file? Actually, if I add both your claims together, I shouldn't be able to tell 44.1 mp3 from a 24/96khz original file, right?

So, the reason this debate continues is simple--the tests to "prove" one way or another are never performed (and I include myself)...so many factors--Cowboy is mixing on an analog desk. This makes the impact lower than someone working ITB. Someone working with a lot of sample based VIs won't hear significant difference because the VIs are mostly 44. Although, I will point out that even most of those are RECORDED higher and brought to deliverable 44.1 during the editing process. That's not trivial to the discussion. Anyway, before I get way off my point--who is going to run these tests? I don't have the capability--which is basically to have as many capture systems as you have sample rates to test--and all using the same AD conversion/clocks. Once they're captured, someone has to agree to the "lesser of evils" in mixing "scientifically" (sic) because You can't use one to mix and simply swap the files in and out--whichever one you were listening to to make all the decisions will be "best". So, who funds such a test? I've actually though about talking Blackbird into participating--as I'm quite sure they could come up with enough IO and HD systems to facilitate.

We have to ask why it should be run? Who stands to benefit? Or LOSE benefit? There's not really money being made pushing double sample rate. I mean--audio isn't a blip on iNTel's radar. I wouldn't think even UA's Sharc DSP chip makers care. There's actually more money to be made if 48khz emerged as "winner"...as that's where converter design matters, IME--I've found anything sounds like a variation on good at double rate...and nothing sounds wonderful at 44.1....48khz, the converter becomes the determining factor. The two things high sample rates "cost" is CPU and HD. Mass consumables that outstripped 48khz production years ago.

So let's say the test is run tomorrow. Everyone can freely download the files....play with them...analyze them...whatever...and make a decision. (for the sake of discussion no one can hear better than 48khz) Does that change our MO? Kinda depends on why we are using whatever we're using. I would change because I'd simply go "hmm...guess I was wrong...." and probably curse myself for ditching my old hardware system that was only 48khz at full capability." But, Cowboy admits double rate sounds better...and uses 44 anyway. He'll still use 44 for whatever reason he does now. You walk around talking about how you were right all along...but, don't change MO. So, I'm the only one changing in that scenario. Now, let's say it's obvious 96khz beats it. Now do you change? Cowboy?

Simply saying that when you look at who benefits financially and what real change would happen...it's hard to find a sponsor...and unless one has Blackbird (or similarly capable studio) at their disposal....one needs one to run something like this. I mean you could drive huge numbers for some period to a website, I suppose...but, that wouldn't continue. It would just be a reference people made I would think after it kind of died down.