Poll: 48kHz vs 96kHz

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

sampling rate????????

  • 64 kHz ("I am a madman who enjoys equipment in pure suffering")

    Votes: 0 0.0%

  • Total voters
    41

Mr. Afternoon

Mixing Engineer & Artist
Joined
May 10, 2021
Messages
1,596
Location
Your Speakers
Let's say John is making a song.
John records all his instruments.
He sits a chimpanzee in front of the computer to do the mixing and mastering.

Great! The song is complete! But wait, should he save space by delivering in 48kHz?
John doesn't see many ultrasonic frequencies in the song. Did the chimp do something? No, it's just the nature of the song.

Should he deliver in 96kHz or 48kHz? Pretend the original recording format is some random crazy high arbitrary value or take the original recording format out of the equation.
 
As its the bit resolution and the convertors linearity (i.e. is it actually 24-bit accurate with equal sized bit steps) which are more important than the sampling rate giving higher frequencies, we're not bats. The higher sample rates are done for technical reasons as it is easier to produce slower roll-off filters which can (for some people) improve the 'sound'. An awful lot was recorded at 48kHz sample rate (and often taken to 96kHz using interpolation in the final offering, which nobody notices). All that said, I usually record stuff at 96kHz/24-bits, being an electronic engineer I like more samples!
 
Yeah, more of a concern when recording. (For all the reasons DuncanS just typed out.) Any sample rate can preserve the audio after the fact as long as the conversion uses something modern-ish like SOX or r8brain. Consumers with muttled sounding low pass filter sporting DACs can upsample as a workaround. (Many consumer AVR products do this by default.) This never had anything to do with preserving artifacts above the audio band.

48k is the sweet spot for low latency needing live sound work. Lowest latency with the lowest processing hit. (96k cuts that latency in half again but with more than 2x the processing.) The default Dante network spec is 48k. Lots of reasons to not be an idiot here!

The DA and AD in my Apogee units sounds every bit as good at SD sample rates as HD. (One of the selling points.)

I'll record and produce at 96k just because the switch goes there and it passively avoids the need for any low pass eq. As long as it doesn't get in the way! And I set myself up with a grossly a overpowered computer for audio so nothing gets in the way! 48k for live sound and recording though. Or if someone has some Roland drum machines and sequencing keyboards they want me to record that only run at 44.1k, that clean 1:1 digital signal coming over the USB ports at 44.1k is looking pretty good!


I'll just add that I don't mean to dismiss any audiophile pursuit of reproducing a full mix with 1:1 HD sample rates and strictly no conversions and all that. I might still agree. I might have made the argument before myself! Just that at the point of recording and production, the mix that is going to be made is 99.99999999999999999999% responsible for anything you might hear. Even if the raw track was converted back and forth to mp3 7 times! Knowing this makes it kind of hard to argue some of this. A full program final mix is another animal though and requires precision to preserve. Vs. say the raw bass di track and raw floor tom track. Keeping everything 24 bit is more important than sample rate.

There were a couple older sample rate conversion apps (past 15 years ago) that mutilated the audio very obviously. People used these in trading circles. I think one of them was called Golden Wave or similar. I think this stuff was wholly responsible for the knee jerk avoidance of sample rate conversion. And then a few circa 1990s consumer DACs that had to be heard to be believed (ie how bad they were!) I've heard some on the fly downsampling that was pretty bad too.
 
Should he deliver in 96kHz or 48kHz? Pretend the original recording format is some random crazy high arbitrary value or take the original recording format out of the equation.

Literally doesn't matter to humans, as you point out. So don't care. Both of them extend well-past the audible spectrum into the ultrasonic - the same file mastered down to 96khz versus 48khz is provably indistinguishable unless you're doing some really goofy resampling stuff, or have a very, very badly designed DAC from 20 years ago. Nobody has ever been able to ABX a difference.

The only reason I want high sample rates is because I want something with the least generational loss from the original digital master as possible.

All things being equal, I'd rather you give me the original file and let me downsample if I want, rather than doing it for me. But that's more of an emotional preference than a quantifiable one.
 
This is an interesting subject. Random ideas.

I currently have two AVP processors in the system, One limited to 48Khz, another capable of 96kHz. Both have access to the same 24/96 files delivered from the same media player. I cant tell the native 96kHz files from the down sampled 48kHz files. At least nothing stands out.

Isn't there a view that some consumer DACs are slightly better at certain sample rates?

If your AVR/AVP is going to down sample everything to apply DSP anyway, does it matter? Is the album mixer's down sample to 48kHz any better than my AVP's?

I've always just felt that, if there is a choice, why wouldn't I opt for 96kHz. if available, But i have started to question it lately.

On the other hand. disc storage cost is so low now. Maybe not so much for pros, but for consumers, prices have dropped.
 
Last edited:
There were a couple older sample rate conversion apps (past 15 years ago) that mutilated the audio very obviously. People used these in trading circles. I think one of them was called Golden Wave or similar. I think this stuff was wholly responsible for the knee jerk avoidance of sample rate conversion. And then a few circa 1990s consumer DACs that had to be heard to be believed (ie how bad they were!) I've heard some on the fly downsampling that was pretty bad too.

I used to use Gold Wave a lot. Back then I was recording at 44.1kHz 16 bit so never/rarely had to resample. It is an excellent program with lots of features and it is very low cost only $19.00. In fact you could use the full program for a period of time without purchasing it. I don't recall if I ever paid anything for using it!

https://www.goldwave.com/
It used to be said that if you were recording to make a CD that you should record at a multiple of 44.1kHz (88.2) rather than 96kHz due to rounding errors in the conversion process. I've never noticed any problem with converting from any sample frequency to any other. So I always record at the highest sample frequency available to me (currently 192kHz), then downsample if necessary.


Keeping everything 24 bit is more important than sample rate.
I agree with that!
 
For delivery? 48k is fine with me.
I've always just felt that, if there is a choice, why wouldn't I opt for 96kHz. if available, But i have started to question it lately.
Yea, that nagging doubt is a SOB.
But since my objective brain side tells me there's no audible difference between the two, I'm going more for the 48k side. My (all SSD) storage is now at 8TB with about 1TB left unfilled, and I'd rather not have to purchase more any time soon. LOL
 
I like 96kHz for the simple fact that my studio monitors have a reported frequency response of up to 50kHz, so I'll get most of that ultrasonic information played back, if it's present. We can't hear it—but it's there in the real world and can affect what we do hear, so with the appropriate end-to-end bandwidth, you can get a more true-to-life presentation. Even higher rates may be useful for recordings going through heavy digital manipulation during mixing, but for consumer playback, bandwidth up to 48kHz (the Nyquist frequency for 96kHz sampling) is entirely sufficient.

And even if you don't care about ultrasonics, a higher sample rate means the anti-aliasing and reconstruction filters can be less steep while preserving all recorded content up to 20kHz, improving phase response linearity within the audible spectrum. 48kHz is already about "twice as good" as 44.1kHz because it has twice the frequency headroom—24kHz Nyquist minus 20kHz equals 4kHz, versus 22.05kHz Nyquist minus 20kHz equals 2.05kHz—but 96kHz is in an entirely different league with 28kHz (!) of headroom. Well-designed filters for 96kHz sampling can preserve recorded content well above 20kHz while still having completely negligible effects on phase linearity.
 
Last edited:
So in essence you are saying that beside the ultrasonic response aspect, 96kHz processes "better"? In general or only with specific DACs ?
 
This is an interesting subject. Random ideas.

I currently have two AVP processors in the system, One limited to 48Khz, another capable of 96kHz. Both have access to the same 24/96 files delivered from the same media player. I cant tell the native 96kHz files from the down sampled 48kHz files. At least nothing stands out.

Isn't there a view that some consumer DACs are slightly better at certain sample rates?

If your AVR/AVP is going to down sample everything to apply DSP anyway, does it matter? Is the album mixer's down sample to 48kHz any better than my AVP's?

I've always just felt that, if there is a choice, why wouldn't I opt for 96kHz. if available, But i have started to question it lately.

On the other hand. disc storage cost is so low now. Maybe not so much for pros, but for consumers, prices have dropped.

I got over 110TB of on line NAS storage in my LAN.

Been recording my LPs at 24/94 since the late 90s.

Now then, I have changed my cartridges twice since and updated my Linn since... so.... yeah.... I can hear those changes in my recordings clearly. So, pretty much I got to redo them, or, more likely, I just shut up and play the records. I ought to note I have also upgraded my AD/DACs.... hmm...

But yeah, with 24/96 you can hear pretty much hear a much better analog to the record. It could be the filters, it could be the better electronics of the AD/DACs as I move forward...

But, why not? The switch is there, the storage is moot, when I record with it, I can not hear the difference between the Linn and the RME. It also sounds very good with Tidal.

I should note, BTW, that ALL of my DACs support 24/96 at the very least.
 
For delivery? 48k is fine with me.

Yea, that nagging doubt is a SOB.
But since my objective brain side tells me there's no audible difference between the two, I'm going more for the 48k side. My (all SSD) storage is now at 8TB with about 1TB left unfilled, and I'd rather not have to purchase more any time soon. LOL

SSD is not really good for storage like that. You will wear it out.

You are much better off with a RAID5 of HDDs... after all, 8TB is nothing, my WD Reds are 10TB each! I just upgraded two of my NAS... so now I'm running 110TB of online NAS at home.

I only use SSDs as the system drives in the user host machines.
 
I like 96kHz for the simple fact that my studio monitors have a reported frequency response of up to 50kHz, so I'll get most of that ultrasonic information played back, if it's present. We can't hear it—but it's there in the real world and can affect what we do hear,

Usually negatively, if at all.
 
Usually negatively, if at all.
So the sound of a violin played live, unamplified, is worse than one recorded digitally with limited bandwidth? I guess that's subjective, but inarguably, the non-bandlimited live performance is more "natural." A higher sample rate will get you closer to that live performance, assuming everything in the chain has bandwidth to match.
 
We get what we are offered for sale as a finished product. Either buy a title or don’t. What is important with digital, is one is given the information of what is being offered before buying.

If one is looking at the issue from an artist point of view, it depends what your computer can do when you keep adding tracks. I prefer 24 bit - 96 kHz for no particular reason, except that I can. Art and how it is presented is subjective to an individual’s tastes.

The “Lord of the Rings” soundtrack DVD-A was at 24-48. It is, or was an industry standard at the time. As an artist, if you believe 24-48 is good, than just do that. But the average consumer probably might expect 24-96. But really, overall, the main thing is just get the titles out there. We still have some of our greatest and favorite Quad titles as DTS-CDs from the late 1990s.

Audio recording keeps upping its game. So, we shouldn’t be surprised if much higher bit rates and kHz are the norm in the future. Better DAC’s. Because technology will be able to do it easily enough.
 
Back
Top