How important is 96k?

  • Thread starter Thread starter kuthu
  • Start date Start date
K

kuthu

New member
How important is 96k? D/A filters in Firestation

So if I'm recording acoustic work, voice, guitar,
drums, flute, banjo, etc. How important is having a sampling rate of 96k? I also thought about getting the Firestation, but it's 44.1 or 46k. If 96k is really pointless, then are the D/A filters in the firestation good enough that it'll produce decent sound?

Thanks,
-J

Here are all the specs on what I'm recording and what I've got:
http://www.homerecording.com/bbs/showthread.php?threadid=116644
 
Last edited:
There is an awful lot of hot air being blown around on this topic. I have yet to hear of a successful pro or prosumer who uses 96K. The consensus seems to be that 96K is more of a marketing gimmic to get more money out of you than it is a technology necessary to be a success.
 
Ahh, so it's like having 16 million colors on a computer screen when the human eye can only differentiate about 800,000 different shades at a maximum. Even though this is true, we all still go for more. Good to know. Thanks.

-J
 
I suggest you do a search on google, as the answer is really too long to put here, but it is NOT a marketing gimmick. I think after 96hz, the human ear would have a hard time differentiating, but I can EASILY hear the differnce in 96hz than in 44.1. BTW - pro setups run at 192 now.

You really have to understand how sampling works to understand why.
 
True - Even with less-than-ideal monitoring, an A/B of a good recording between the 40 range and the 90 range is pretty obvious.

That being said, when I'm going for a high sample rate and I know that the finished product is a CD, I tend to use 88.2kHz (the multiples thing).

THAT being said, YES. Good, even great quality recordings CAN be made at 44.1kHz. I'd keep at 24 bit or higher as long as possible though...

John Scrip - www.massivemastering.com
 
If you think about sampling a 22Khz wave at 44Khz. It means you only get 2 samples to describe each wave cycle. From those 2 numbers, the D/A converter must try to reconstitute the original wave form.

Sampling at 96Khz results in 4 samples per wave cycle giving the converter a much better chance of accurately recreating the wave
 
massive master do you think that recording at 88 is a better idea than 96 if its going to be on a CD? Its because of the dithering/mixing down to 16/44 right?

Could you explain more?

Thanks... i have the ability to track at up to 96 and i was planning on doing this but....hmm.
 
human ear hears at a max of around 20 -22 kilhertz. the nyquest therum of sampling (done a while ago) says that if we want to get a good representation of someting we must sample at 2 times the highest frequency to avoid aliasing. hence 2 x 20000 herts = 40 000 herts, a cd is 44100 hertz. The good people at (was it sony?) decided that 44100 samples would be enough to cover the range of human hearing.
What they did not consider was that music / note/ whatever is usally made up of a lot of harmonics which have a higher frequency. Sampling at a higher frequency therefore catches more of these high harmonics.

Is that right? correction are definately welcome. music 207 coming into play here!
 
The way I understand it is like this:

Bitrate - how often you take a sample of the music
Bitdepth - how much precision you have in that sample

If you want to think about music as a stream of numbers flowing accross your computer screen. If you have a high bitrate you will look at the screen 100 times a second (for example) if you have a high bitdepth you will be able to see the number that is scrolling by to the 16th decimal place (for example)

If you don't sample enough times then you won't get an accurate picture of the numbers. If you don't sample with enough precision, you won't get an accurate picture of the numbers.

They work very closely together.

* This is how my mind things of these things. I realize this subject is much more complex than this, but I feel this is an accurate representation of what is going on*
 
Yeah, that's a pretty good analogy. If you imagine a sinusoidal waveform (i.e. a squiggle that goes up and down in regular peaks and troughs) then the resolution (i.e. 16-, 24-bit) is the vertical axis. The higher the resolution, the more points you have on the axis (e.g. 256, 65k, 16.7 million). The converter rounds the amplitude of the waveform off to the nearest point on the axis, so if you have loads of points close together you'll record something closer to the original.

The sample rate is the number of times a second the converter makes a rounding-off measurement like this. 44.1kHz is 44100 measurements a second.

The D/A converter then does this in reverse to play back - it puts all these points on a new graph and then essentially joins up the dots.

All of which leaves me with the question still - if I have a Delta 44 capable of 96kHz, does is matter if I sample at 48? I'm not short of hard disk, btw. But I would think that it is more important to have 24-bit resolution if I did have to make a choice.
 
Recording at 96k for final delivery at 44.1k will introduce a few problems. The conversion is not perfect because the 96 does not easliy divide into 44.1. This introduces artifacts to the recording. These artifacts are not that noticible as distortion but it will drive you crazy when you bounce your perfect mix then listen back and it sounds different?!?! Try bouncing something at 96k to 44.1k you will hear that is sounds different when bounced. I also use the multiples idea and record at 88.2k because I think the bounces sound more consistant with my mixes.

I also record at 24Bit because of the reduced noise floor.
 
that makes sense, i'll try at 88.2 from now and see how that works.
 
photoresistor said:
massive master do you think that recording at 88 is a better idea than 96 if its going to be on a CD? Its because of the dithering/mixing down to 16/44 right?

Could you explain more?

Thanks... i have the ability to track at up to 96 and i was planning on doing this but....hmm.
Click to expand...

Honestly, I really can't tell the difference. It's come up in the "brainstorming" sessions of "audio scientists" that when you're SR is up that high, you're best off working in a multiple of the FINAL SR. I know it's not as simple as "dropping every other sample" but somehow it does make some warped sort of sense...

I really don't feel that it makes enough of a difference to make it "Law" though... 88.2 never really caught on anyway.

As far as I'm concerned, once you hit 44.1, I'm much more concerned about bit depth than sample rate.
 
96k isn't VERY important but differences are more noticeable when staying in the computer. i guess the computations are more accurate because it seems at least to me that the final product is smoother and more open on top and the mids are more defined and not muddy (96/88.2)....at least for how i work...YMMV
 
that's weird though because there's some scientist's Law that says you need to sample at twice the highest frequency. we can't hear higher than 22kHz so 44.1 should be fine (which is I guess why it was chosen for Red Book CD).
 
quote> that's weird though because there's some scientist's Law that says you need to sample at twice the highest frequency. we can't hear higher than 22kHz so 44.1 should be fine (which is I guess why it was chosen for Red Book CD).
------------------------

Putting the marketing "hype" aside..... yes, Red Book standard WAS chosen rather wisely at 44.1 kHz as that allowed for sounds up to 20kHz with roughly 2,000 Hz of spare overhead for the anti-aliasing filter to roll-off the high end smoothly (too steep an analogue filter introduces phase distortions which is why they don't just use a brick wall filter at 22kHz).

However, that was then.... and this is now.

RedBook was chosen, wisely, based on the the technology of the time, particularly in regards to storage capabilities. CD's at the time were still "cutting edge" and holding more data on optical disc was still only experimental at the time.... so there was no way that they could have sampled at 48kHz, nevermind 96kHz, because then the CD-Audio discs just would not hold enough music to be commercially viable products. Any storage medium has to be able to hold 10 - 20 tracks of a typical song (i.e. a full album) otherwise its back to the drawing board.

If they could have chosen 48KHz at the time of RedBook, they would have (actually.... thats also debateable... since even back then they had no idea what we'd be doing nowadays with this 96 and 196kHz recording).... but CD's just couldn't hold that much info... so we got stuck with 44.1 and 16-bit-linear as the final product which was good enough as a "consumer" end-product.

Most importantly.... just because 44.1 makes a suitable "end-product" good enough for consumer listening... does not mean that the guys who designed RedBook thought that studios should master in 44.1kHz and 16 bit! down-converting for final product consumption is fine in reality..... but if your masterwing and mixing involves lots of outboard equipment bouncing and lots of going in and out of the computer... then you want to try to use the highest sample and bit rate reasonable so that you minimize all the damage dony by the endless Analog--> digital> ---> analog -----> digital -----> analog -----> digital conversions. Every A/D/A conversion you do adds distortion... so it can't hurt to use the best A/D/A converters and use the highest sampling rates.
 
kuthu said:
Ahh, so it's like having 16 million colors on a computer screen when the human eye can only differentiate about 800,000 different shades at a maximum. Even though this is true, we all still go for more. Good to know. Thanks.

-J
Click to expand...

that could be one way to look at it; but do this. put a nice 3d background on your desktop. put it on 16bit then on 32bit. you'll notice the shades be not so choppy. now this may not be the same with audio, but i know my equip is capable of 96kbps just in case. ;)
 
Industrial - good post. So would you reccomend sampling at the highest rate I can? Do you have any experience of the dithering problems some people have said they've noticed?

I wasn't suggesting recording at 44.1/16, just pondering why Teacher reckoned increasing the sample rate improved his top end when it really shouldn't make any difference in theory...?
 
Industrial - good post. So would you reccomend sampling at the highest rate I can? Do you have any experience of the dithering problems some people have said they've noticed?

I wasn't suggesting recording at 44.1/16, just pondering why Teacher reckoned increasing the sample rate improved his top end when it really shouldn't make any difference in theory...?
------------------------------

Industrial replies:

To be honest, I don't have "golden ears"... not many people do... I do tend to extremely notice small differences at higher sampling and quantizing-rates (bit depths).... but I guess some people can hear it better to others... I figure, just because I can't hear it, doesnt mean their lying or they are wrong...

Also, the ear is something that can actually be "trained" to some extent... this is something most sound engineers can tell you... It's not that music guys have better hearing than the general population.... but I guess with experience, you can actually learn what different sounds and artifacts are in the mix, and you can determine what piece of equipment put it there. For instance, I can't tell you what a dithering error sounds like because it's never really been pointed out to me... but having lots of experience with compressors and noise gates, I can easily pick up other peoples processing errors just by listening to the compression and the clicks and other artifacts.

But if you measure the signal with analyzing equipment, the difference is obviously there... whether it can be differentiated by a given individual, or not.

-------------

Errors incurred by dithering, is something I have had a hard time differentiating.... so I can't speak on that matter...

However, if you are not aware there are actually several algorithms which one can use for dithering (i.e. Rectangular, Triangular, Guassian, Shaped Triangle, Shaped Guassian)... Not all editing software has these dithering choices... but each of them.... theoretically.... has different characteristics. Rectangular dithering provides the lowest noise at the expense of distortion whereas Guassian may provide more noise but less distortion... or something like that...

I haven't tried all the algorithms yet, so I don't know if the difference is really perceptible...

I suppose the lower the sampling rate is, the more the dithering would become apparent to even those who don't have "golden ears".
 
Eventhough I am not sold on the 96K thing yet, I do agree on the damage done by the ...digital--->analog---->digital---->analog---> digital conversion thing that seems to happen, unless people have extremely hi-end converters and equipment/compressors worthy of putting your signal back to analog. Usually a few stages of low-end D/A and A/D converters can kill your sound. For hobbyists and home studios....Once digital...Keep it digital!
 
You must log in or register to reply here.
Back
Top