I've been recording in 24 bit 44.1 and I've never had someone say hey I think it will sound better in 48k.. I'd rather have a nice stable environment then a pubic hair difference in audio quality

 

Audio quality and tech specs can't be decoupled was the point I was trying to make. But fair enough.

 

It might come down to “noise floor.”

Bits are all about dynamic range. Hertz are all about the width of the frequency spectrum.

 

I record everything at 96 and work the mix and master at 44.1. While some prefer to work on 48 and convert to 44.1 at the very end, i try to get the highest resolution on the recording process to avoid convertions on the latter process. I don't like 48 in any manner. Most of the plugins nowdays have oversampling, so i don't feel i need to upsample the whole session in order to get sound quality. I mean, 48 is 'standard', i don't want to enter the rabbit hole of this discussion. Just sharing my experience.

 

And do you think you can say the same thing about 24bit vs 16bit?

 

The easiest approach for anyone to follow would simply be to understand what your end-goal is any time you begin a new project.

Is your project going to end up on CD? 16/44.1 is your answer.
Is your project going to end up on DVD/BluRay? 24/48 is your answer.
Is your project going to end up on Vinyl? 24/96 (This is absolutely ridiculous and illogical, but hey, it is what it is!)

 

Actually it's more of a weasel hole.

 

Either way, it's a hole

 

Interesting, where can we read this article?

 

The Sense and Nonsense of High Sampling Rates
https://digital-audio-systems.com/sinn-und-unsinn-von-hohen-sampleraten/
https://digital-audio-systems.com/sinn-und-unsinn-hoher-samplingraten-prox/?lang=en

 

You may want to check source material on these.. There's some snake oil being peddled there...
From that article; "This means that a sound with 60 dB is not six times louder than a sound with 10 dB, but has 600 times the intensity. As a rule of thumb, a difference of 10 dB is perceived as double or half the volume."

The above sounded like hogwash to me so I looked around to verify what I have always known to be fact and sure enough; "This means that the two sound intensity levels differ by 3.01 dB, or about 3 dB, as advertised. Note that because only the ratio I2/I1 is given (and not the actual intensities), this result is true for any intensities that differ by a factor of two. For example, a 56.0 dB sound is twice as intense as a 53.0 dB sound, a 97.0 dB sound is half as intense as a 100 dB sound, and so on."

the above (in yellow) was quoted from;
https://phys.libretexts.org/Bookshelves/College_Physics/College_Physics_1e_(OpenStax)/17:_Physics_of_Hearing/17.03:_Sound_Intensity_and_Sound_Level

±3db changes in volume/intensity are perceived to be "half or double in volume or intensity" not 10db. Personally I would take that "digital audio systems" (quoted in red above) thing and chuck it where the the sun don't shine...

then again this is all from "the internet" - where it's possible to prove that the earth is flat and the sun rotates around it

 

Are you the author of this article?

 

Hi @mondomorte, no I am not the author of the article. Please read the following post.

 

Last edited: Mar 9, 2025 at 9:38 AM

Thanks @saccamano for investigating the veracity of this. I will no longer recommend this article.

Double sampling rates (by Dipl.-Ing. Thomas Sandmann)

Audio CD / Professional DAT recorder = 44.1 kHz
Consumer DAT recorder / internal frequency of many effects units = 48 kHz
Some modern effects units now work with double sampling rates = 88.2 to 96 kHz

According to Shannon's sampling theorem, the highest audio frequency that can be displayed is half the sampling frequency. At 44.1 kHz, the audio signal could theoretically reach up to 22.5 kHz. If frequencies higher than half the sampling frequency (also called Nyquist frequency) occur in the audio signal, non-linear distortions occur, which are referred to as aliasing (from “alisas”, Latin for “the other”).

It must be ensured that no such frequencies are present in the spectrum before the A/D conversion.
Every A/D converter therefore includes an anti-aliasing filter which, due to its finite edge steepness, must begin attenuation before the value of 22.05 kHz. If we assume that the audible audio range extends up to 20 kHz and should remain completely unaffected, the requirement for the ideal filter is such that the slope only starts above 20 kHz and has fallen so far by 22.05 kHz that levels in the frequency range above this are negligibly small.

Until recently, it was almost impossible to design such a filter satisfactorily. Either the filter was designed with a very steep slope, so that the attenuation only began above 20 kHz, or a flatter slope was chosen and the attenuation began well below 20 kHz. In the first case, the filter produced a distortion due to a high ripple in the audible range; in the second case, the distortion resulted from the filter slope itself. Only the more recent developments of digital filters show results in which the requirements for a wider passband and low residual ripple are solved equally satisfactorily.

Conclusion:

From the point of view of the aliasing problem, we are well served with today's standard sampling frequencies, and the increase to 96 kHz hardly brings any improvements. however, some processing, especially digital equalizers, benefit very clearly from the 96 kHz format.

In an effects unit of the highest quality, these processes should therefore be carried out at double the sampling rate. After that, however, there is not much to be said against converting back to 44.1 kHz. Whether 96 kHz makes sense in every effects unit remains questionable. In any case, increasing the resolution to 24 bits is more important than the higher sampling rate.