Best Sample Rate

It is a very long time since I read about this stuff so bare with me ... Just tried to find some good old documents to share with you, but no luck so far (thousands of files laying around).

First of all I believe there will be more rounding "errors" converting 48 kHz to 44.1. The computer operate with integers. But nothing to concern about anyway I believe. Just think I remember converting huge files and had an experience of it taiking longer converting 48 kHz than 88.2 to 44.1. And I think I believe I read it some time long ago. if you have the chance, use e.g Adobe Audition and 10 minute long file you generate and convert it. I am not 100 % on this.

A digital equalizer works very diffrent than an analog one. With a low samplerate there is no chance for the eq to be exact or precise. Let say you try to tune in exact at 440 Hz, then you will always touch frequensies nearby. This effect is lesser and lesser noticable at higher samplerates. And Nyqist has something to do with thus too ... Can't explain all the mumbo jumbo right now cause it is long time since I was dealing with it. The thing I remember and always have in mind is that stuff about bit is important regarding tweaking dynamics, and dealing detail with EQ etc. then samplerate is important.

I will look some more around for documentation. There is plenty of it around, and people do not always agree 🙂

Absolutely, and thanks for your reply on this.



I am just figuring that the extended frequency range being sent in to a Dynamics controller would naturally effect the detection of how it compresses the audible frequencies. And of course if the signal coming in does not extend past the audible frequency range then there would not be any detection thus no changes in the detection and or compression being applied. But, from my experience in mixing and mastering at higher sample rates, even with unconverted 44100 to 96k stems my mix is profoundly more natural sounding compared to the lower sample rate mixing scenarios...



Everyone keeps mentioning to just use your ears, but it does not seem that black and white to me. There is obviously a scientific reason and or advantage in using higher sample rates when multi track mixing and or signal processing. I also think that in most situations, anyone seeking the audible difference when using a higher sample rate are then most likely not using the higher sample rate in a way that would lend better sounding results. On my end, I seem to be using it properly due to the overall increase in instrument detail, RMS weight / headroom, and of course the separation of sounds from one another in the mix is much more detailed....

dogwhistles...

Bee Jay wrote:
"Increase in headroom" or loudness because you're using a higher sample rate?

Sorry, you've lost me... How are the two related?

Yeah, I agree... I don't see how sample rate is going to affect headroom or loudness. It does make sense, though, that mixing at a higher sample rate (even if the original recordings are simply upsampled) could sound better... at the very least, any aliasing introduced by the mixing/processing steps would occur at higher frequencies, so less likely to generate audible artifacts.

James

lugic wrote:
First of all I believe there will be more rounding "errors" converting 48 kHz to 44.1. The computer operate with integers.

Read up on sample rate conversion--the rounding error comparison is incorrect. And Logic doesn't work with integers, by the way... it's a floating-point mix engine...

A digital equalizer works very diffrent than an analog one. With a low samplerate there is no chance for the eq to be exact or precise. Let say you try to tune in exact at 440 Hz, then you will always touch frequensies nearby. This effect is lesser and lesser noticable at higher samplerates.

You're talking about bandwidth (Q), which is independent of sampling rate. And it's no problem to perform phase-linear eq at lower sampling rates. Some eqs may sound better at higher sampling rates, but not for the reasons you mention.

And Nyqist has something to do with thus too ... Can't explain all the mumbo jumbo right now cause it is long time since I was dealing with it.

Seriously, man, no offense, but this kind of "explanation" does more harm than good to the audio community. If you're citing real math... great! Or if you're citing real world experience... great! Otherwise, what you're offering is... gotta love the Internet... 🙂

James

Yes. It's an example of something that is common - a basic understanding of digital audio stemming from the late eighties when we really didn't know how to design "good" or optimal digital audio systems - poor jitter, bad convertor, poor signal-to-noise ratios, trying hard to record towards 0dBFS, no dithering, fixed point low resolution sampling, magazine articles about sampling using the familiar "stair-step" illustrations etc etc.

There are a lot of persistent digital audio misconceptions, including the obvious ones such as trying to record hot, or digital audio being "stair-stepped" and the more resolution you have, and the more samples, the "finer" the steps and hence the better "quality". And plenty more besides.

That's why I drop into threads like these and try and spread some more up-to-date info. Hopefully before repeating the spectacular highs of the "Levels in Logic" thread... 😉

Yeah, I agree... I don't see how sample rate is going to affect headroom or loudness.

In fact, I can make an argument for (potentially) decreasing the headroom at higher sample rates (the opposite of what the poster stated.

Let's say your audio content has a lot of energy in those higher "inaudible" frequencies. In a lower sample rate system, that stuff is filtered out - no problem.

When you up the sample rate, all of a sudden that high frequency info is now contributing to the overall levels - they will get metered louder (although you won't hear much difference back in the real world) and therefore your headroom will decrease.

However, with around +1500dB of headroom in a well-implemented floating point mix engine, it's not going to make much practical difference in terms of headroom. 1500dB is more than enough - whatever your sample rate..! 😉

Yes it is old 🙂 hxxp://sysdoc.doors.ch/TERRATEC/2496-en.pdf

mattrichardofyork wrote:
Hmmm so no simple answer to this one then lol

Actually, I think there is a simple answer:

Reasons to use 44.1k:

The Good Reason: Higher sampling rates hog disk and CPU and provide only minimal sonic improvements that may not be audible in project studios or with typical listening gear, iPods, etc.

The Dumb Reason: Nyquist-Shannon proves that CD quality sound is perfect. (Hint: no, it doesn't)

Reasons to use higher sampling rates:

The Good Reason: Given your particular equipment chain, you may hear a sonic improvement you think is worth the extra disk and CPU.

The Dumb Reason: The more samples, the smoother the waveform, the more dynamic range, sound is warmer, etc. (Hint: none of these things are true)

James

lugic wrote:
First of all I believe there will be more rounding "errors" converting 48 kHz to 44.1.

absulutely irrilevant and not audible with Logic 8 on Leopard Core audio

check out here:
http://src.infinitewave.ca/

this test are made from indipendent engeeners

the resultin is without doubt

*Logic 8 and Leopard digital conversion from 96 to 44.1 shows absolutely THE BEST RESULT* (no artifacts)
Pro tools is a bullsh*t if compared 🙂

That is why so many engeeners suggest to never use conversion
*they works on PRO TOOLSSSSSS!!!*

LOGIC IS THE BEST ONE!

In any case
I remand that the CD is not the unique support

is you are sure that don't need to make a MTV video ov your song you can use 44.1

BUT IF YOU WANT obtain the professional master using 48 *pro standard* of Digital Audio from more tha 25 Years!|

Leopard and Logic are the best
in any case mastering is required before make a CD...

Mastering studios prefer to works with 48... 96 or 192 KHZ!
(because they have analog Solid State preanps and High quality converter for make the conversion!

but I repeat Logic8 and Leopard Core Audio are on the TOP OF THE DIGITAL CONVERSION
no artifacts at ALL!

I suggest to use 48Khz on logic 8 and Leopard core audio!
G

check this canadian guys
http://www.infinitewave.ca/

you will be able to understand why Logic is the most advanced DAW...

🙂

Since so many sophisticated answers have been given, i must add another one:

The answer to this question mainly depends on which species you belong to: Human or Bat

Cheers,

says the Fox

Samples Rate is Mathematics...
this question is not related with

belong to: Human or Bat

the answer must objective...
is an objective answer and there's absolutely nothing of subjective in Samples rate numbers!
this leads to false answers and urban legends

this is objective
http://src.infinitewave.ca/

you are free to use 44.1 for your music
I prefer 48Khz for processing power reason...

Because I'm not sure if my Music goes to CDs only...
I'm the owner of original Master...
I must have the better quality of it in my archive for future events

this is a objective answer that are focused to the question of the Topic....

Hope that it will be helpful for the Logic users!

G