Yes 32 bit. Then smash those masters into a sausage.
I have experimented extensively with 44.1kHz, 48kHz, 96kHz, and even 192kHz. I have recorded and produced entire projects at all these sample rates. When it comes to the recording itself, I notice very little difference between the different sample rates… However:
I think some “processing” or FX do benefit from higher sample rates. This is more noticeable with some plugins than others. Also, some plugins automatically upsample to 192kHz for processing which poses a string of questions: 1) Do you think resampling is entirely without artifacts, distortion, etc? There are many sample rate converters that claim superior sample rate conversion. You can see some comparisons here: http://src.infinitewave.ca/
2) If resampling “does” affect audio in some way, doesn’t it make sense that if plugins are upsampling and downsampling from 192kHz, that by running your project at 192kHz natively, you could avoid that extra sample rate conversion entirely in the processing? Is it possible this is a benefit? I think it’s an interesting question at least.
In regard to higher sample rates… Do you think how “steep” the slope is on antialiasing filters in converters could potentially have audible effects? If so, the filter can be moved far above the audible range if working at 192kHz. At 44.1kHz, the filter must be VERY VERY VERY steep to avoid antialiasing. It’s possible that could affect the sound.
Anyway, I think the most important thing is to just use your ears if you have the means to test it. Then you can decide for yourself without needing to ask. Learning how to trust your own ears is one of the most important things you can learn as an engineer. Getting an answer from someone on here and just doing what you’re told is probably not the best solution. After many many experiments with this, I feel there is indeed a very small benefit to running at higher sample rates. The difference is small though. I believe you can produce completely professional sounding results even when working at 44.1kHz - there are FAR more important issues than the sample rate. Still, I do prefer working at higher sample rates myself just for that tiny added benefit.
Lastly, I know it’s slightly off subject, but have you heard DSD? I have a Korg MR-1000 DSD recorder that I have compared directly with my converters (Apogee DA-16X). The Korg can operate at 5.6mHz. Yes, that is MEGAhertz, not kilohertz. How does it sound? UNBELIEVABLE. It’s kind of shocking when you set up an A/B comparison between PCM audio and the Korg DSD. Many people act as if digital is perfect as it is and there is no point in anything besides 44.1 kHz 16-bit audio. For the public, that may be true - even mp3 seems ok for the consuming public. For audio engineers, if you can make something better, and you actually hear it as better, would you prefer to ignore it, or use it to your benefit?
The Korg MR1000 is a one bit system, so 5.6MHz equates to stereo 24 bit at 192KHz - it’s just another pointlessly high sampling rate for marketing purposes.
A normal 16 bit/44.1KHz CD runs at 1.411MHz (as data is read serially 1 bit at a time).
As has been mentioned - your ears are easily fooled by what your eyes expect to hear.
So it’s all pointless then?? If your going to mix down to 441 then just stick to it in the first place right?
Spent 20 plus years in audiology and after testing thousands of people both normal and impaired, most drop off after 4k or 8k and even then most audiometers won’t measure over 10k. Then, theres the difference between hearing level and sound pressure level and brings in things like equal loudness contour.
Yes. That’s very reasonable. I think so. In the Bon Jovi camp, the know it 
OFFLINE processing benefits from higher BIT RATES.
Online processing (mixdown/DAW playback) are operating in the 32-bit floating point realm, no matter the bit they were recorded at.
Some plugins don’t upsample (mostly old bad ones), and may add in noticeable artefacts. Upsampled plugins are benefitial when doing OFFLINE processing, not so much in realtime processing.
And remember the plugin upsampling (4x 8x etc) is upsampling the BIT RATE (for better Dynamic Range/Signal to Noise ratio), not Sample Rate.
So next, Sample Rate has nothing to do with Dynamic Range what so ever (that’s bits remember). Sample Rate is all about frequencies. Read up on Nyquists theory.
And remember: There are NO mechanics inside our auditory system that can convert frequencies above 20kHz into electric signals (to be sent to our brain via our nerves). Hence there are NOTHING to be interpreted by our brain, aka heard.
You just can’t combat nature.
If you are hearing things, no matter what it is, it HAS to be within the human hearing range. Agree? Which is?
Sample Rate is all about frequencies. Nothing more, nothing less. That’s the beauty of it.
Don’t make the easy parts of this more difficult than it is.
Sample Rate is all about how to capture and play back high frequencies according to the Nyquist Theory.
Sample Rate : 2 = Upper Frequency captured and played back perfectly (what comes in, comes out. Exactly the same).
And as long as you are over the (mechanically) possible human hearing range, you are all good.
PS. If you are not to abuse your preciously recorded sound sources with crazy amounts of OFFLINE processing.
Don’t make these topics (Bit Rate/Sample Rates) more difficult than it is:
Sample Rate = Frequencies
Bit Rate = Dynamic Range
Have you heard the korg then? You know it’s only marketing because you have done comparisons? I’m willing to bet you’ve never even heard it… Hehe.
Cubase processes at 32-bit, yes. It is not only Offline processing that benefits though. Realtime processing benefits as well.
WRONG - there are some very good plugins that don’t upsample. WRONG about only being beneficial for offline processing. Realtime processing benefits as well.
WRONG. Many plugins upsample the sample rate ALSO. Universal Audio does this with many of the plugins for the UAD card (up to 192kHz). Do you think there could be a benefit to running natively at 192kHz so you can avoid this upsampling and downsampling? If not, WHY not?
Mostly true - although how steep the antialiasing filter is might affect the sound. At 44.1kHz, they have to use an EXTREMELY sharp filter.
Also, some plugins benefit from processing at higher rates.
Yes, Cubase processes in the 32-bit realm when realtime mixing/processing.
That’s why I didn’t mention in particular that it was benefitial, you can’t do anything about it
Yes, ALL processing will benefit from the 32-floating point processing. But the realtime will be done in that realm no matter what. NOT so with OFFLINE processing, so just take precautions 
Not familiar with UAD, but many manufacturers are using many terms at will.
To better the Dynamic Range, or expand the useful range for calculations, you have to upsample the bits involved (to keep the quantization errors way out of our hearings range).
For OFFLINE processing you don’t need to upsample the Sample Rate (the speed it works at), the calculations will be done anyway, But…
…But finally, here is where the REALTIME processing gain the most. To upsample the rate/speed it works at. Nothing to with the sound per se, but the DAW/computer will cope better on big mixes/projects in real time (calculate faster).
So while many things discussed here is done automatically within the DAW, I have for the most part pointed at things we can choose to change ourself. Whether this can be heard or not within the human hearings range.
But I guess we can agree on this:
Bit Rate = Dynamic Range
Sample Rate = Frequency/speed
The easy part is what goes on inside our DAW automatically. The hard part is things we can change and manipulate our self, as most people tends to think to hard about these things.
I’ll leave it there
Yes, I have used them (MR1 as well), though it must have been 6 or 7 years ago when they came out.
1 bit DACs are not quality, they are cheap to make, which is why they were used in walkmans and cheap CD players. What ever sampling frequency or bit level you record at with them, they will not achieve a better dynamic range than a 16-bit CD and probably no better distortion than a good quality tape recorder. 1-bit DACS have high distortion in the upper frequency range.
Though Korg cleverly used the 1-bit DAC technology concept to market the product range. They will record at 24-bit/192KHz, but never achieve any better quality than 96dB dynamic range (that’s 16-bit level).
Well put. As if to illustrate just how crazy these discussions can get (especially amongst unsuspecting members of the public who don’t have degrees in digital signal processing), Neil Young has been touting a $400 “walkman” to allow people to listen to 24-bit 192kHz files and then goes and records his latest work on a 1947 “Voice-O-Graph” – the novelty recording booth seen in old movies where you can record yourself directly onto a disc.
From an artistic perspective it’s valid, of course, but I would feel a bit sad for anyone who might splash out their hard-earned cash to buy his device and then listen to intentionally lo-fi recordings on it.
My advice to any newbies here or anyone interested in sample rates:
Don’t listen to guys like Andyjh who act like they have all the answers. Just do the tests yourself and judge based on what YOU hear. Trust your own judgement. Take the time to experiment and actually test the difference. That is how you grow as an audio engineer - and not by blindly following the opinions of others. I will tell you right now though, that I believe you can make a pro product with or without higher sample rates - even though I very slightly prefer higher rates… VERY slightly. Test it and find out for yourself what YOU think.
Thank you
Thank you
Thank you
Thank you
Thank you
Thank you
Thank you
Thank you
Thank you
and a major ‘Mahalo’!
The best plug ever made.
Yer ears!
And you can’t go out and BUY better ones.
Yet! (perhaps one day)
Sending much Aloha.
{‘-’}
Made a few experiments and always returned back to 44.1/24. Best performance/sound ratio for me.
Never was sure if I really hear or just imagine any improvements in sound/definition/clarity at higher rates. With real audio recordings it’s not so easy to make an A-Z comparison. You’d need two identical stacks of gear, fed by a signal splitter on the way in. Up-/downsampling after recording is not be the same.
Saying that, I use a lot of upsampling UAD plugins and believe there is a point in using higher rates at ‘strategic points’ in the signal chain.
I said I leave it her, but
Sample Rate is about speed/frequency (how often).
There is the advantage for using upsampled (sample rate, not bits) plugins. They perform the calculations X time faster, making the “calculation cue” move faster. May or may not be an issue, but for heavy loaded sessions it probably help the cpu’s doing their math to their fullest potential.
PS. While at it. That’s why we have lower latency when tracking through our DAW in 96k vs 44.1. This is also the only advantage IMO (ref. the human hearing range explaination earlier).
When doing offline processing, it doesn’t matter. Then our computer can in a steady tempo calculate whatever math equation it has to deal with… And it will take a few ms more
That was the Sample Rate part. The Bit Rate part is for another day.
Man, this is hard to explain as easy as possible (after years of reading thousands of pages on the topic)
Man, this is hard to discuss in an easy way (with the lot that have not used years reading thousand of pages on the topic) 
Now I leave it with that. That’s a promise
44.1 kHz, 24 bits.
No one can hear above 17 kHz, taking the Nyquist theorem and modern filter technology into the equation gives us more than enough with 44.1 kHz of sampling.
A bat and a dog will disagree here, but we produce music for the homo sapiens species, so 48, 88.1, 96 and more kHz are just a marketing fad, esoterics. Same scam as the EUR 5.000,- digital audio cables for audiophiles. 
But 24 bits is absolutely preferred, not because it sounds better (it doesn’t), but because it makes life easier. You can record with insane headroom and don’t have to take care about the hotness of your signals.
32 bits (floating point) sounds good to me, too, but I’m not absolutely sure if audio interfaces give you proper 32 bits during recording (with values outside of [-1; 1]) or if they truncate (which would be the logical thing to do), so this is for all the stages after the recording only.
However:
TEMPORARY (!) upsampling makes perfect sense for some types of audio processes, this includes some types of synthesizers - and has strictly mathematical reasons. But the main audio path should always be as close as possible to 2x Nyquist frequency of the human ear (plus some frequency “headroom” for the reconstruction / aliasing filter).
Here’s a primer on why I’m right and why the “zomg 192 kHz” - guys are wrong:
Please be aware that marketing people will always find ways to sell you their esoteric stuff, I even heard utter stupidity like “improved phase correlation and clarity because of extra harmonic content”, which makes no sense at all, because the human ear simply can’t pickup those higher frequencies in the first place.
When you go for anything above 44.1 kHz you are only wasting CPU cycles for nothing.
As directed in my first post in this thread
Again, this is pure nature, physics and math. Don’t fight it, accept it, move on, and go make some music