Truth or Myth #2 “96Khz is better then 48Khz”

[bteck]

Truth or Myth #2 “96KHz is better then 48KHz” and “196KHz is better then 96KHz”

This is TRUE

Sampling rate of your DAW determines the frequency bandwidth of the system. That is given Fs , Fs/2 is the maximum bandwidth called the Nyquist frequency.

In the 70’s and 80’s as Rupert Neve was building some of the finest solid state mixers it was knows and reported widely that his Neve mixers had extended frequency bandwidth . Common those days was bandwidth of 20Khz or 30 KHz where the Neve mixers went up to 100Khz. Report had reported on many locations that many of the users of these mixers said that they “sparkle” as opposed to lower bandwidth mixers.

There are many other technical reasons why working with higher sampling rates are good. One for example is it allows plug-ins to do a better job, in an EQ for example when you are applying equalization that affects the high part of the audible spectrum, the filter that affects the audio will be “warped” by the Nyquist frequency. As a result there is modification to the filter that was not to be there. When the Nyquist frequency point is increased ( i.e. the sampling rate is increased) then this no longer becomes an issue. Other benefits of higher sampling rates are that it allows plug-ins to minimize audible distortion and other noises which are common in all digital processors.

So given SNR and distortion stay the same, 96 is better then 48 and 196 is better the 96.

The real question is how important is it for the type of work and production you are doing?

Working at higher sampling rate is very expensive. The equipment cost more money, you get less tracks and need a lot more memory and DSP power. Depending on the process, most plug ins will require twice the processing and twice the memory for every doubling in sampling rate. Some plug-ins require 4 times the memory and processing for the 2 X jump! That means a process at 192 will require 8 times the processing then at 48!!

In my opinion, the quality increase is apparent although you get diminishing returns, that is the improvement from 48 to 96 is much better then the improvement from 96 to 192. In many cases I find it is much better to work at 48 as the track count is larger and that will do more for the project then sampling rate. It all depends on what you are doing and what you want to achieve.

[Haigbabe]

Truth or Myth #2 “96KHz is better then 48KHz” and “196KHz is better then 96KHz”

This is TRUE

So given SNR and distortion stay the same, 96 is better then 48 and 196 is better the 96.

Well I'm going to do a 160 degree turn and head straight back to the music store and get a 196k(sic) converter.

Haigbabe

[kmshroom]

Well I'm going to do a 160 degree turn and head straight back to the music store and get a 196k(sic) converter.

Haigbabe

heh heh.

[Phil O'Keefe]

http://www.musicplayer.com/cgi-bin/u...&f=3&submit=Go

Somewhere over there on that forum is a huge thread where the whole 44.1 / 96 KHz thing was argued to death...

Anyway, I DO hear a little something extra going on at higher sample rates... but to me, it's far less than the increased sonics that result from recording with the same Fs at 24 bits instead of 16. And I agree - what to use will depend on the project. I'm working on a Christmas CD where it's primarily harp and vocals... a perfect candidate for a higher Fs... but I'm still doing it at 44.1. Why not 96 KHz? I've got the tools, and it's not like it's gonna burn up all that much HDD space (limited amount of tracks - it's a simple project)... but I'm staying at 44.1 KHz simply because it's going to CD, and it will NEVER wind up on a DVD-A release. And it sounds just fine as is. I might have decided differently for another project, but this is working fine on this one.

But I definitely agree - DSP is better at higher sample rates.

[Phil O'Keefe]

Here's the actual link to the 96 KHz thread over on George Massenburg's forum:

http://www.musicplayer.com/ubb/ultim...=000822#000000

Bring reading glasses - it's 33 pages in length!

[skip b]

I do know that Bob Katz upsamples to 96khz or mabe now 192khz to make his edits, addition of compression,eq,etc. then SRC back to 44.1 then dither.So I think if these guys do it, there must be something to it.I hear a big difference at 192khz. Some say that the only reason people use higher frequencies is to compensate for poor converters.Well if thats true why would companys like WEISS,Z-SYSTEMS,etc have the capabilities in their equipment.Surely their converters are top notch.

[froyo]

Hello. Actually some of the same people (mainly Nika Aldrich) have discussed this here on the DUC. Here's one of the latest from last month and it's only 3 pages (depends on your individual settings for the DUC). Basically this falls mostly under the category of Myth #2. A properly designed A/D and D/A system (i.e. converter design) will be at its most accurate at 44.1 Khz. But in order for converters to be properly designed, the price point will increase. So depending on the converters you are using you probably will find differences. Human hearing tops out mostly at between 18Khz and 20Khz. So 44.1 Khz is more than plenty to accurately capture that range. Higher sampling rates deals with frequency we can't hear. Analog gear deals with audio differently and it is subject to intermodulation distortion and things like that maybe what is being referring to with the Neve gear.

All converters do have some advantages for higher rates. These were also discussed by Chris Townsend from Digidesign and Nika. As it turns out generally most EQ's do not benefit from higher sampling rates, but compressors do.

Please check these links as they have all the info on them that there is no space to write on this post, replete with very specific explanations as to why the things I mentioned above are what they are.

Sampling Rates
Going from 96Khz to 44.1 Khz
192 Khz

[bteck]

Sorry I had a small error, I wrote "That means a process at 192 will require 8 times the processing then at 48!!
"

It is 16 times !! ( not 8 times ) for tools like linear phase filters.

Also EVERY digital EQ working at 44.1 or 48 has a warping problem at the high end. There have been tools like RenEq that try to compensate the filters for it, but still the problem exists in all digtal EQ working at low sampling rates.

[kmshroom]

I do know that Bob Katz upsamples to 96khz or mabe now 192khz to make his edits, addition of compression,eq,etc. then SRC back to 44.1 then dither.So I think if these guys do it, there must be something to it.I hear a big difference at 192khz. Some say that the only reason people use higher frequencies is to compensate for poor converters.Well if thats true why would companys like WEISS,Z-SYSTEMS,etc have the capabilities in their equipment.Surely their converters are top notch.

the reason Bob Katz does this is because when you upsample something that has already been recorded as digital in order to process it with compression, etc, the ugly aliasing that is a by product of most plug-in processors resides in the inaudible high frequencies, and so when we listen to it, it sounds cleaner, because the ugly stuff is too high up there. then it gets chopped off when it goes down to 44k for a CD, so it's definitely gone.

that's sort of how Bob Katz describes it in his Mastering Audio book, anyways. he has all sorts of graphs in it to illustrate it all very well.

[qbert1]

This may have been covered in one of the other threads, but I thought I'd share what I've learned as a musician in a study on intonation...

One thing a higher rate does for you is capturing what's known as "resultant tones". These occur when two or more frequencies are played at the same time. The math is pretty simple if 1000hz and 750 hz is played perfectly in tune by a pair of instruments (I'm know these don't line up to real notes, I'm just trying to keep the math simple) their wave collisions create a third tone of 250 hz or two octaves below 1000hz (1000-750=250).
Now, in cheaper pipe organ designs, manufacturers were able to exploit this principal and save room and money by producing their lowest sounds with two pipes out of the range of human hearing. 40hz could be produced by 26000 and 25860.
How does this help us? I believe reverbs are where the warmth and sparkle of resultant tones really come in to play. Al around us, there exist many higher frequencies than we're usually aware of. If we don't capture these frequencies on a recording or produce them with a 96khs reverb process, then they will never have a chance to interact. But as stated before, is the extra money worth the subtle shimmer and warmth gained by resultant tones? Up to you I guess.