44.1 v 96 kHz

bassdude · Post by **bassdude** » Sun Oct 12, 2003 5:28 pm

Think about mixing and processing these waveforms together.. every process changes the sound more far away from original analog counterpart.

While I understand the principle of mixing at 96kHz and above, if you have processes that are drastically changing the original waveform that is not part of the actual intended effect of the process then that is just poor programming/algorythms. Once it's passed the the converters it's all one's and zero's.

astroman · Post by **astroman** » Sun Oct 12, 2003 8:37 pm

of course there are differences in programming quality, but it's really not that simple with all those 'ones and zeroes'

You probably remember the math tortures from school with all that sine and cosine stuff ? Unfortunately the value of those functions frequently approaches (and in theory even reaches) zero and infinity rather frequently.
The most horrible situation for any computer math

With a tiny deviation (the slightest inaccurancy) in calculation you can yield a tremendous effect. To pick the proper functions is located somewhere between art and voodoo.
Possibly one of the reasons why our Sharcs sound so good

cheers, Tom

ps: tnx for the compliment, GaryB

Music Manic · Post by **Music Manic** » Tue Oct 14, 2003 3:42 am

That's why the more bits the better because when the math divides and multiplies over the highest possible value it will truncate data corrupting the signal.No algorythm can deal with that only the bit rate.
Digtal is just a representation of the signal just like the dots on the TV.24bit is what we need to get the ear to believe the sound is true,but we haven't succeeded in the LSB yet as things go.
Truely believe there's different directions to come in digital.

astroman · Post by **astroman** » Tue Oct 14, 2003 7:19 am

this is non-linear math, in the critical 'parts of the curve' a few bits more don't help at all.
A good 16 bit algorithm can still superceed a bad 24 bit one in sound quality.

Something isn't perceived as more 'real' because of a higher resolution or increased frequency response, but due to proper phase information - as far as that's possible with current equipment.

Phase means nothing but a different distance in the original soundsource. A guitar or a piano produce various sounds at various parts of the instrument. You can neither record that wave pattern, nor reproduce it with a couple of loudspeakers.

The impression of realism increases the more you're able to fake the original image.
The ear is much more sensitive to that part of the signal because it is so much more important for survival. The frequncy response only adds to this. The higher the frequency the better you can locate a source.

For example there are those in-ear mics which deliver a stunning impression of moving objects if played back via headphones.
Or police sirens and street noise in hip-hop records, which often make you believe that's happening right behind you when listening to that stuff in a car.

cheers, Tom

valis · Post by **valis** » Tue Oct 14, 2003 8:11 am

Its true that the errors in conversion (and processing) can drastically affect phase as well as frequency, I think this is one reason everyone's pushing to 96k and higher.

I would think the real reason companies are pushing for higher clockrate is because it would make the cost of the converters cheaper. Higher sampling rate means that the errors in the waveform would deviate less in cheap converters, therefore allowing you more latitude in cutting corners before people can discern the audible effects. For example, almost all of creative's cards exhibit massive problems in their cheap converters that result in horrible phase errors. Roll out the 96k (audigyII) and voila, same cheap crap with an apparent increase in quality.

Or am I mistaken?

Post by **garyb** » Tue Oct 14, 2003 11:28 am

Stige · Post by **Stige** » Tue Oct 14, 2003 4:34 pm

Valis, my thoughts too.
At higher rates the sound is closer to what it should be in analog domain. This is not all about just how high our ears can go.

Music Manic · Post by **Music Manic** » Wed Oct 15, 2003 6:59 am

So if you use a good convertor plus high sampling rate?

Music Manic · Post by **Music Manic** » Wed Oct 15, 2003 7:07 am

Also referring to AEL.It uses a method that mixes the analogue signal of the source and adds it to the digital conversion to give a better sound.
This is done with a analogue circuit so I can't see why it can't be mixed with the conversion from Pulsar to create an analogue sound.Interesting,I think we are going back to come forward.

Nestor · Post by **Nestor** » Thu Oct 16, 2003 10:23 am

To my experience, there is no point in using higher sample rates, I nevertheless would always use as high bit rates as posible, cos the difference IS truly enormous. I enjoy leastening something at 24 or even 32bit floting bits very much, it sounds different!

In those matters, I trust above all, my ears, not the theory.

The difference you can get from going to higher sample rates is so extremely small, that it does not worth the effort. It is better to use the power of your computer and the space of your drives storage, recording 24 bits.

I also try to do as little processing as posible once I've finished recording the audio tracks... I choose what I want, and try to make as many shortcuts as posible to get the sound, without much processing. If posible, no EQ for instance. The result: a much better sound.

Sample rate is far away in importance compared to many other factors which are related to mastering. Panning for instance, it's very important and you can without any complicated anhancement, give to your mix a truly different, professional sound. Remember that, sa many already said in here, perceptions is in fact the matter...

Do all your processing in 24 bitas thow and you'll get important benefits out of it.

Again, try to overload as little as posible your audio tracks with unnecesary FXs, also, DO NOT touch much your final mix. Try to reach a soon as posible, your required sound, not going throu too many plugins and if posible, reaching the required sound without using ANY EQs.

When recording, everything is best done at it's source rather than traying to solve the problem or searching a different sound later.

I even am starging recording voices directly with some EQ tweaks, compressors and reverb, direct, getting incredibly proffesional results.

<font size=-1>[ This Message was edited by: Nestor on 2003-10-16 11:37 ]</font>

valis · Post by **valis** » Thu Oct 16, 2003 11:43 am

Again unless I'm mistaken, there's more to 96k (and etc) than 48khz high frequencies. Since the audio rate is moving at a much higher speed, many more audible frequency divisions can be reproduced.

If 48khz can only produce 375hz and 425hz with minimal aliasing (i'm choosing arbitrary values here for argument's sake, please don't tell me math is wrong

) then 96khz is able to do 375hz, 400hz and 425hz resulting in more resolution within the bandwidth of our ears.

Now I'll admit that everything else Nestor says is excellent advice, and I myself work in 44.1 for the time being...but I suspect that ther IS a benefit provided the a/d and d/a are quality designs. Do an unaliased osc test in 44.1 or 48khz sometime and listen to the frequencies jump as the scale moves up (RedMuze once illustrated this to me).

Post by **garyb** » Thu Oct 16, 2003 12:18 pm

of course you're right(again),but consider this,there are recordings currently at 44.1k that are stunning.better is always better,but as i said earlier,let's not be fanatical about it.use what makes the most sense,all things considered(the final outcome/use,playback abilities,economy of storage and cost vs. returns,etc).

Rob van Berkel · Post by **Rob van Berkel** » Thu Oct 16, 2003 1:50 pm

Having read all those wise postings, I remain with one crucial question: Why record and process at 96 when the result has to be downsampled to 44.1?
Is there one simple answer?
Cheers,
Rob

valis · Post by **valis** » Thu Oct 16, 2003 8:36 pm

88.2 would be a better choice. I think most of us are content to work in 44.1 for the time being unless we're working on a project specifically for a high resolution format. And I completely agree with you as well garyb...for now

Immanuel · Post by **Immanuel** » Sun Oct 19, 2003 11:41 am

On 2003-10-14 08:19, astroman wrote:
The higher the frequency the better you can locate a source.

To a sertain limit, where i starts to become dificult to localise a sound again (IMO).

astroman · Post by **astroman** » Sun Oct 19, 2003 12:12 pm

I guess that's because the reflections of the sound become more prominent, too.
The ear is a great construction, probably that's why the frequency response is not linear, so the effect you noticed is compensated

cheers, Tom

at0m · Post by **at0m** » Sun Oct 19, 2003 1:28 pm

On 2003-10-16 14:50, Rob van Berkel wrote:
I remain with one crucial question: Why record and process at 96 when the result has to be downsampled to 44.1?
Is there one simple answer?

Lets say you have to make 1000 calculations and you can work with numbers from 0-9. Wouldn't your final result be better if you could use big numbers or decimals, even if you'd round off the result?

my 2cents.

Joop Nijenhuis · Post by **Joop Nijenhuis** » Tue Nov 25, 2003 8:41 pm

On 2003-10-19 14:28, at0mic wrote:
On 2003-10-16 14:50, Rob van Berkel wrote:
I remain with one crucial question: Why record and process at 96 when the result has to be downsampled to 44.1?
Is there one simple answer?
Lets say you have to make 1000 calculations and you can work with numbers from 0-9. Wouldn't your final result be better if you could use big numbers or decimals, even if you'd round off the result?

my 2cents.

In my opinion you have to look what source you have and what the destination is. If the source is an other digital system than it makes no sense in recording at a higher rate than the source. The destination is in most cases a cd. This means that if you decide to go for the 96Khz you have to reformat the recording to 44.1 in order to get it on cd. However the proces of reformating is that troublesome that the gain of recording at 96kHz is totaly lost (you can compare it with the soundblaster trouble). For those about eq and other editing. The more pro systems all do internal calculations on a much wider foot in order to keep the calculations as pure as possible. I record the more critical naturesounds and we feared MD, DCC, Mpeg x. In practice there is hardly a difference. Most heard errors are own fault by handling the mic in the wrong way or just too much noise (wind, trafic, airplane, rain..). Music isn't that demanding, we tried classic quitars on a Paris system with only a 442 interface. This doesn't go any faster than 48Khz with a resolution of 20bit. Going down to 44.1Khz didn't reveal any difference.