96k

  • Thread starter Thread starter JuSumPilgrim
  • Start date Start date
J

JuSumPilgrim

New member
96k guys
If you understand digital audio then you know about the Nyquist limit and what happens if you record something over the Nyquist limit without a suitable filter. If you record 28kHz, you will hear a tone at 16kHz during playback.

The same sort of thing happens when capacitors can't handle the high frequencies. They discharge at the wrong time and add peaks where there weren't any in the audio.

As I said in my 96k column, I had this happen on the new Steely Dan mixes, and Tom Jung ran into the same problem with a Sony bitstream demonstration when using an amplifier that was not band limited to 20kHz.

This does not happen every time, bu since you can not directly hear the contents way up there, then oscillations above the 20k limit in the analog console can adversely effect the audio below it.

When digital audio first started being used widely in the early 80s, there was all of this low frequency farting that happened with digital machines connected to Neve consoles. The digital recording process was blamed. It turned out that Neve consoles have low frequency noise around 10 to 16 Hz that would screw with the DC removal circuits in the digital machines causing the problem. You can tell if a Neve has this problem by looking at the phase meter. If the phase meter is flopping back and forth when no audio is passing through the console, then there is low frequency capacitor noise, period. The phase meter should come to rest dead center if the console is quiet. Still in the year 2000 a good 30% of the studios I work in have not fixed the problem. Their answer is "Nobody complains when recording analog" or "We don't hear anything, so it is not a problem."

1) For multitrack recording I think 96k is a waste of time and storage. At best, 96k is double the resolution in the time domain with the penalty being twice the storage requirements.

Maybe there is a slight difference for mixing that some people think they hear, so fine... use it for mixes, but nobody has been able to show me that playing back 24-48 tracks of 96k sounds any different than 24-48 tracks of 48k.

2) 24 bits, on the other hand, is easily heard, is 256 times the resolution in the level domain, and only requires 50% more storage.

I rest my case

Roger



Just something I came across at Roger's forum I thought Id share.


JuSumPilgrim
 
How many people here record in 24/96? I've never really done it, and I'm wondering if any of you guys can honestly tell the difference after you've dithered down to 16/44?
 
Good point i have a 24/96 card and inever record above 16/48 16/44.1

I tried at 24/48 once and windows wouldnt play it (not even XP)
and it confused my soundcard!!
 
I now mixdown to 24/88.2... I have always heard a difference in word-size (and always mixed-down at 24-bit), but only recently, it became very clear to me how much the detail (such as reverb tails, or transients) sharpens considerably at the higher resolution for some material.

Bruce
 
Last edited:
I'd dearly love to experiment with the higher res and I'm sure there's a difference. Unfortunetly, I'm limited to 48KHz by my ADATs, Motu and GenX6. Personally, I don't hear a difference on my sysytem between 44.1 and 48k but I do between 16 and 24 bit. Now if I was doing solo acoustic or classical work, there might be advantage in 88.2 or 96k in clarity and the lower track counts of these styles would offset the storage/processing hit on a computer system. I'd also venture a guess that given the past strides in computer and hardware/software development that in three years or so 24/96 will be the norm and the these types of discussions will be based around the virtues of 128bit/2GHz recording (look man, I can HEAR light!!).
 
I have never done any A/B testing so I will not say anything about what I prefer. But whatever mr Nyquist has to say I have to assume a lot of phase information has to be lost in the higher frequency content even if the amplitude, on average, is correct.
So I guess I would use 88.2K if I was AB-micing but stay at 44.1 if I was XY-micing or multitracking. 48k and 96k seem somewhat stupid to me.

By the way, don't we use anti-aliasing filters any more?
 
Oh, a lot of things! But in this case that the sampling frequency needs to be at least twice the highest frequency component of the signal you wish to reproduce. Nothing you need to be a genius to conclude, really...
 
Which is where the anti-alias filter comes in.
 
i always record at 24bit/44.1khz due to system/resources limitations....whenever i upgrade ill take a crack at 24/88.2........
 
I also record at 24/44.1. To my ears there's a pretty noticable difference between 24 and 16 bit, even after 24 is dithered down to 16.
 
BasPer: "I have never done any A/B testing so I will not say anything about what I prefer. But whatever mr Nyquist has to say I have to assume a lot of phase information has to be lost in the higher frequency content even if the amplitude, on average, is correct.
So I guess I would use 88.2K if I was AB-micing but stay at 44.1 if I was XY-micing or multitracking. 48k and 96k seem somewhat stupid to me."


What are you talking about?? Why are 48k and 96k stupid? I really tried to follow your logic, honestly.
 
>Why are 48k and 96k stupid?

Oh, I just think the conversion from 48/96->44.1 is going to be a lot more difficult than from 88.2->44.1
 
so what you are trying to say is that the math involved when dithering from 88.2 to 44.1 is a lot simpler than from 96 to 44.1, yielding better results?
 
lucid,

i did read 88.2 was easier and yielded better results somewhere here on the BBS but i cant for the life of me remember who said it.....
 
Back
Top