How Big a Difference from 48khz to 96khz?

In your opinion, how much of an audible difference is there between 24bit/48khz audio and 24bit/96khz audio?

Both of the audio interfaces I've bought and tried so far can only go up to 48khz (the Cakewalk UA-1G can record at 96, but you can't play audio at the same time... so you can record, but won't hear anything...). Is the difference from 48 to 96 big enough to be worth upgrading to another interface that can do 96?

Your insight is much appreciated. Thanks!

Really? Hmmm. I was under the impression that the difference was bigger than that. Say, like the difference from 16 bit to 24 bit.

What about up to like, 192khz? Is that a noticeable jump? (I dunno if many people record that high)


And no, I don't play much classical, mostly hard rock/metal (but what if I were to throw together some orchestral stuff with synths n whatnot?).

Some thoughts:

More than 85% of what I do is 'classical'  (Actually mostly baroque but I digress) the difference between 16 bits and 24 bits is agnostic to genre - As CJ points out - there are massive S/N and dynamic range dividends in recording at 24 bit - there is little reason to record at 16 if you have the storage and processing power.

WRT 48Khz or 96Khz - that is project dependent - If my work is destined for radio broadcast 48Khz is fine as a starting point.  Even my film work has been completely 48Khz. Some of my minimal mic orchestral stuff has been at 96khz - however the moment I do multiple spot micing on sections (woodwinds, strings, continuo, etc) then the benefits of 96khz fall apart for my ears. I feel that 96khz provides great image stability when doing a great true stereo recording - there are other reasons for 96khz but that is at the top of my list - 

Working on an orchestral session right now  - gotta go...

-D 

CJaysMusic

 

and you can tell the differenc ebetween a 16bit audio file and a 24bit audio file?
You would be the only one in the world!
Cj
EDIT:

.

 

Actually the club is larger than you think my friend.  Being able to tell the difference between a capture and mix at 16 bits compared to 24 bits is done with little effort.  As the signal chain improves so do the benefits of the additional byte.  Back to my baroque orchestra - I'll post some samples saturday morning (they'll all be 24 bit :)

-D

I accidentally recorded a song in 16 bit. It's on my website, the only song there, called 'My Best'. After all the compression and limiting, you can hear the noise, but I only noticed the noise once I realised it wasn't 24 bit (a year after finishing it). Before then I thought the song was pretty good in terms of noise levels. I'm going to re-record certain sections when I make my album though to get a better performance and to significantly reduce this noise.

The benefit in 24 is simply to get the digital noise floor lower than the equipment noise. If your acoustic environment is very noisy and your signal chain is noisy too, you won't get much out of 24 bit recording. But in a quiet environment with quiet gear, you WILL notice the difference, but probably only AFTER compression and limiting. You will also probably notice it if your track is very dynamic and you turn it up loud (such as classical).



As for 48 vs 96, I personally think it's a load of bull. If your filters are good enough (and according to Bob Katz, that's not to difficult to achieve) then there is no noticeable difference in blind testing by professional audiophiles.

I am a novice in this whole digital audio recording world, however I do have years as a computer hack and photographer.

My take on it is, that while the FINAL output is hard to differentiate between high and medium bit rates, there is a reason to capture at the higher rates (48K vs 16k for instance ).  Whenever you manipulate a file, compress it, add effects, etc there is a certain amount of rounding that goes on in the computer, this is the nature of doing advanced math on 32 or 64 bit numbers. 

The higher the capture rate, the more data you have per second and the less you will notice sound degradation from that rounding.  The real question is at what point does the degradation become audiable.  I probably could not tell, but my ears are trashed from my days in the military and from years of standing too close to a drum kit, but for someone with good hearing I would bet they can detect the difference.

Again, I am new to audio so feel free to question this assumtion, however I do know on photographs and video it does make a marked difference.

Regards, -Harry

Mostly, (I think) the reason some can hear a difference is that they expect to hear a difference.  Me?  I'm old and well, my hearing isn't as good as a teenager but still pretty good

I think we tend to get hung up on the technical stuff and ultimately feel that somehow better gear will make better music.  (and to be fair it does make it 'nicer' not necessarily 'better').

I don't think I've ever done anything in SONAR at anything other than 16/44.    That being said, I know I'll get some heat, but really today's listening 'room' is a pair of earbuds, a car system that is seriously not flat (with the two 15" subs in the trunk) with the added road noise and most listen in an MP3 format on an phone or other player and well... I digress.

To the OP, I wouldn't think you're missing out on anything.  At best IMO, I would use up to 48/24 and as others have commented, that would satisfy anything remotely plausible in the human range of hearing. 

I think Ethan Winer has a video on this subject.  Google it and it is pretty amazing to hear what you think you hear.

Good Luck!

Of course one use for higher sample rates is reduced latency, where extremely low latencies are essential.

I have pasted this to save time.

"First thing you need to know of is a Buffer.
This is a small piece of memory which allows 2 pieces of software running asynchronously to communicate.
[A bit like a key-stroke buffer - you type letters on the keyboard, these go into a buffer, and later Word (other packages are available !) collects them from the buffer.]

Using unreal (but simple) figures.

Assume a buffer size of 100 samples.
If you have a sample rate of 100Hz (100 samples per second), your latency will be 1 second.
i.e. it takes 1 second for the first sample to make it's way through the 100 sample buffer.
OK?

If you now double your sampling rate to 200Hz, (200 samples per second)can you see how the samples will make their way twice as fast through the buffer? Now the latency (the time it takes to pass through the buffer) is now 0.5 second.

OK, scale these up to 4800kHz, a 1000 sample buffer and we have:-

1000 / 48000 = 20.8ms latency.

Double your sample rate to 96000 and your latency is:-

1000 / 96000 = 10.4ms.

Correct ! 

well, thanks for setting me straight! 

(you can tell that i don't get hung up on the technical stuff -  ha!)

I guess I've not suffered from many latency issues - but I do use mostly outside synths, etc.
 
But back to the OP question.  Can you hear a difference? 
 
I can't.
 
<edit for spelling>

Just my 2 cents... dunno if CJ was referring to the below example of 16 vs 24 bit but:

I sample a lot of instruments (one hit - with articulations) to create virtual instruments for use within Kontakt.. 

I used to sample at 16 bit and I thought that it sounded pretty good... but about 2 years ago, I tried to sample at 24 bits and I will say that I can hear a difference in my sample libraries... the 24 bit stuff sounds a lot smoother than the 16 bit stuff... sorta took the digital edge off the sample libraries.

Now, perhaps I imagined this but I have been sampling at 24 bits since.

Jim

I have 24 bit DVD-A's that sound really good.. better than 16 bit cd's ?? I think so.

A couple of thoughts:

1. If you run things at 96kHz and feel you are hearing a difference, don't automatically assume that the sample rate is the only thing that changed (some people will even compare different converters at different sampling rates and claim the difference they're hearing is due to the sample rate itself).

There's a lot going on in modern converters - sampling method, analog electronics (including anti-aliasing or reconstruction filters), oversampling (including the digital decimation filter when downsampling, or interpolation filter when upsampling), clocking/synchronization.

2. People often confuse what it might be possible to hear in terms of high frequencies under "laboratory conditions" with what kind of (very) high frequency content is actually present and can be reproduced mics/speakers/analog electronics/etc. It's a bit dubious as to whether anyone can hear frequencies much over 20kHz at all - even when played back at high volume. But even if you can, there's no way your ears are as sensitive at high frequencies as they are at 2kHz and, in the real world, very high frequency content is usually not near maximum amplitude. If you can barely hear something (if at all) at -2dB, how do you expect to hear it at -20dB?