96 khz. Does it really matter?

Ray Gianelli

Member
Messages
1,344
While I understand how increased bit rates in digital recording will reduce quantization errors I don't understand why pushing the sampling rate up past 48 khz would make a difference. Does it? 96 khz is way past the Nyquist point.
 

bluestarbass

Member
Messages
78
I think it depends on your converters, but in my 002 i can tell a HUGE difference between 44.1 and 96. I typically record in 88.2. That said I think if I was using a rosetta or aurora, theyre 48 would sound better than my 96 on my 002.
 

Ray Gianelli

Member
Messages
1,344
I think it depends on your converters, but in my 002 i can tell a HUGE difference between 44.1 and 96. I typically record in 88.2. That said I think if I was using a rosetta or aurora, theyre 48 would sound better than my 96 on my 002.

Thanks for the reply. You know, there are some things that don't seem like they'd make a difference on paper but do in the real world. I guess this is one of them.

OTOH, I've got a lot to learn about the mechanics of recording before this will be an issue!
 

loudboy

Member
Messages
27,306
For a home studio? Maybe.

It's about last on the list of things that make a difference, so if you think you can hear it, your computer can handle it and you can deal w/storage and backup, why not?

Loudboy
 

dougb415

Member
Messages
10,478
I tried going from 44.1 to 96 on my PC. My Dell system sputtered and basically would not run. Now I understand that my PC knowledge is slipping, but this is a dual-core unit that I thought should be able to handle at least a couple tracks this way. Nope.
 

Nomadgtr

Member
Messages
413
I can run several tracks at once at 96 on my 3.2Ghz PC and my understanding from the Cakewalk forums is that the Dual and Quad core machines are barely being pushed even with lots of plugins running. You may have a soundcard issue especially if you have other devices that it has to synch with. I have to reset my card to switch back and forth from 44 to 96 otherwise it won't lock with my external RME A/D D/A converter.
 

triple_vee

Senior Member
Messages
1,141
I forgot which guitar rag I read it in, but it said that if your computer could handle it, do it.
 

malabarmusic

Member
Messages
1,702
Assuming you have good converters on both ends and the end goal is "CD quality" (as opposed to MP3), it's probably a good bet. 192, OTOH, is most likely counterproductive.

- DB
 

TAVD

Guitar Player
Platinum Supporting Member
Messages
3,818
I guess it all depends on who you ask. I heard from one designer of high end converters who says they should sound the same at any fsr for properly designed converters. While that may be true, I have little doubt that some converters sound better or worse at a given fsr, in the real world. What do your ears tell you about your converters @ 48khz vs. 96khz? There's your answer.
 
Messages
3,640
I guess it all depends on who you ask. I heard from one designer of high end converters who says they should sound the same at any fsr for properly designed converters. While that may be true, I have little doubt that some converters sound better or worse at a given fsr, in the real world. What do your ears tell you about your converters @ 48khz vs. 96khz? There's your answer.


I'm guessing that was Dan Lavry, and I agree totally....theoretically, there's nothing to be gained by going to 96kHz. But, in practice, and especially with cheaper converters, the higher sampling rates often sound better due to poor anti-alias filter routines/designs.

Use your ears...that's all you can do.

Cheers

Kris

PS: I record at 44.1khz....that's all my DAW supports, and it's never been the limiting factor in my recordings.
 

timmers

Silver Supporting Member
Messages
730
I agree with the "use your ears" comments above. Everything in recording is inter-related; i.e., your ability to hear differences in sample rate is going to be affected by your A/D converters, which is going to be affected by the mic cabling running to them, which is affected by the mic hearing the sound, etc. And then it's all affected by your particular monitoring situation. SO many variables, especially when you're doing it in a less-than-perfectly-controlled environment, as most home recording folks are.

That being said, an engineer friend of mine up in NYC who's a real heavy on the session scene has told me a few times that if forced to choose between sample rate (44.1 vs. 96) and bit depth (16 vs. 24), he'd *always* go with the higher bit depth. Says it's the more audible of the two. Hopefully, gear limitations & availability won't force you to have to make that choice!

T
 

MikeMcK

Gold Supporting Member
Messages
7,090
From what little I've read on it (and this information is a few years old):

a) bit depth is much more important, especially after you've hit the max Nyquist sampling rate for audio (IOW, above ~ 40 kSps)

b) a lot depends on filtering. If your input is accepting analog much higher than the Nyquist limit ( >= 2X the sampling rate) you can get aliasing artifacts in your audio band.

c) ADC quality is crucial... for awhile, the faster ADCs were designed to be faster, not more transparent. This is most likely fixed by now.
 

Tonefish

Senior Member
Messages
5,282
without wanting to get enthralled into a controversy, I just keep going back to my engineering college days when we proved through analysis of laboratory data, that we could get aliasing at the Nyquist limit. We analyzed and calculated to something like 2.76 rather than 2, and that was just with the one signal. I feel better with 10x but I probably won't pay for that.
 

elambo

Member
Messages
2,375
...but 24 bit vs 16 bit stores 256 times as much detail...

Hmmm.. Now that's cheating. It's not really 256 times more "detail" because you're referring to increased volume resolution - amplitude - not the resolution of the waveform's frequency, which is really what matters.
 

elambo

Member
Messages
2,375
without wanting to get enthralled into a controversy, I just keep going back to my engineering college days when we proved through analysis of laboratory data, that we could get aliasing at the Nyquist limit. We analyzed and calculated to something like 2.76 rather than 2, and that was just with the one signal. I feel better with 10x but I probably won't pay for that.

That makes sense.

If there are any non-believers of higher sample rates, listen to a well-recorded SACD.

Personally, I record everything at 48KHz because that's the SR at which I deliver my final product and I know that a SR conversion would be more damaging than the slight gain I'd get from 96. However, I have recorded sessions at both 48 and 96, with a lot of eq and compression, then printed the results from each and the 96 certainly sounded better. But after I converted it to 48 it lost that advantage and wasn't much better, if any, than the sessions that stayed at 48 the entire time.
 

gixxerrock

Member
Messages
4,060
There is a lot more to it than just the Nyquist limit. No question, there are significant algorithmic advantages to oversampling beyond 44 kHz. However for my gear and ears, I am perfectly happy using 24bit 44kHz for everything.

Shawn.
 

Bassomatic

Member
Messages
12,336
But after I converted it to 48 it lost that advantage and wasn't much better, if any, than the sessions that stayed at 48 the entire time.

It would be interesting to perform one of those nulling tests between the two mixes, if they were the same in every other respect.
 

elambo

Member
Messages
2,375
It would be interesting to perform one of those nulling tests between the two mixes, if they were the same in every other respect.

That would be interesting. I didn't try it then and I'm not sure I still have the sessions, but it should be pretty easy to repeat. I was triggering external instruments via MIDI so that they could be sampled at 96 for one session and 48 for the other, then I eq'd and compressed, etc., all in the box. I'll try to put something together.
 

elambo

Member
Messages
2,375
There is a lot more to it than just the Nyquist limit. No question, there are significant algorithmic advantages to oversampling beyond 44 kHz. However for my gear and ears, I am perfectly happy using 24bit 44kHz for everything.

Shawn.

44.1 can sound great. No reason to feel like you have to change.
 




Trending Topics

Top Bottom