48 kHZ or 44.1 kHZ ?

[cron]

It's mostly the last two comments in that thread I was addressing. I don't really see how you can double blind test 44.1 native vs 48-to-44.1 resampled audio without rendering the same project at different sample rates as your first step. If the listener is successfully distinguishing the projects through ABX testing, a null test via phase inversion will reveal the actual differences between the files. I'd be very surprised if the result was only aliasing near Nyquist.

I would be VERY surprised if there was ANY aliasing. ALL current DAWs have very good anti-aliasing filters. There will probably be some difference due to resampling in one side versus direct rendering in the other side, but I doubt they would be noticeable.

That's the thing. I'm suggesting that poshook was able to successfully double-blind test because the difference in plugin behaviour was way more profound than merely better behaviour at Nyquist.

As an example of a huge sample rate clanger, the grain size of Cyclop's Transformer oscillator is directly linked to sample rate. Doubling the sample rate therefore plays grains an octave higher.

Edited my suggestion about null testing due to simulated non-linearities in audio software, but it could still reveal some interesting things if not being a true 'document' of only the difference.

edit: incorrect user name

[poshook]

i wonder as i always thought 44.1 kHZ and 24bit is enough but more and more say especially for sound design, making music/sounds for games etc. that 48 kHZ is a must.

1. is it true?
2. how much impact on the performance (PC/MAC) does it have?

1. 48kHz while mixing and convert to 44.1kHz when bouncing provides much better results compared to 44.1kHz in whole process. The reason is aliasing. The more plugins you use while mixing the more benefits you get from 48kHz against 44.1kHz. I changed from 44.1 to 48kHz year ago and never regret.

2. Performance impact on current CPUs is close to zero

Mixing at 48k will indeed reduce aliasing, compared to 44.1k, but I believe that won't survive the resample process, and will be almost identical to a mix done entirely in 44.1k.

The second that you down sample from 48 to 44.1, you lose any extra benefit.

Try things doing not just believing. I did a lot of blind tests and every single time picked the one mixed in 48k and then downsampling to 44.1k over 44.1k

How would I do this test?

By listening to audible differences? (10000 variables, and the noise is at -90dB or so anyway)
Check a spec analyser? (Which one?)


Basically, using a higher sample rate will move nyquist higher and reduce aliasing. All well and good.

The process of resampling down will essentially reverse this process. Back to where you started.

Hey, that is simply not true. There is no reverse as there is not only final resampling that matters. Try thinking about e.g. 40 audio tracks and some AUX, and 2BUS, each with the EQ and compression etc. Each of those around hundred plugins is processed at 48kHz instead of 44.1kHz (except those with automatic optional oversampling). So the resulted 48kHz file is based on huge number of calculations of all included tracks at 48khz. The downsampling to 44.1kHz is applied as the final phase on already rendered file. That means that this downsampling is not able to take you back you started. Is that clear?

[poshook]

i believe he was not talking about switching to 44.1 on render, but rather rendering at 48k and then resampling to 44.1k. So the render itself is in 48k, just resampled to 44.1. i too think there would be little to none audible differences between them.

That is absolutely true. Sonic difference between 48kHz render and downsampled 44.1kHz one is non-audible. Difference between 48kHz original render and 44.1kHz original render is more obvious and amount of difference depends of the number of processed tracks and number of plugins introducing audible aliasing across the entire mix.

[fmr]

Hey, that is simply not true. There is no reverse as there is not only final resampling that matters. Try thinking about e.g. 40 audio tracks and some AUX, and 2BUS, each with the EQ and compression etc. Each of those around hundred plugins is processed at 48kHz instead of 44.1kHz (except those with automatic optional oversampling). So the resulted 48kHz file is based on huge number of calculations of all included tracks at 48khz. The downsampling to 44.1kHz is applied as the final phase on already rendered file. That means that this downsampling is not able to take you back you started. Is that clear?

Which nowadays are the vast majority of the plug-ins (especially if we are talking about GOOD third party plug-ins). Which means that the rendered audio, no matter if it was in 48 kHz or 44.1 kHz, was already downsampled in the first place. The "huge number of calculations" happens anyway, no matter if you are working at 48 kHz or 44.1 kHz.

I may accept there may be differences, what I doubt is that they ARE noticeable (and if noticeable, if they are enough to pick one over the other as "better"). People waste too much time nowadays over these "picky" questions. If it sounds good, then it IS good

It's funny there are people discussing the merit/demerit of 44.1 kHz vs 48 kHz, while other people route their final mix through an analog stage (or even at several stages during the mix), converting back and forth between digital and analog, with all the consequences that imply

[poshook]

Hey, that is simply not true. There is no reverse as there is not only final resampling that matters. Try thinking about e.g. 40 audio tracks and some AUX, and 2BUS, each with the EQ and compression etc. Each of those around hundred plugins is processed at 48kHz instead of 44.1kHz (except those with automatic optional oversampling). So the resulted 48kHz file is based on huge number of calculations of all included tracks at 48khz. The downsampling to 44.1kHz is applied as the final phase on already rendered file. That means that this downsampling is not able to take you back you started. Is that clear?

Which nowadays are the vast majority of the plug-ins (especially if we are talking about GOOD third party plug-ins). Which means that the rendered audio, no matter if it was in 48 kHz or 44.1 kHz, was already downsampled in the first place. The huge number of calculations happens anyway, no matter if you are working at 48 kHz or 44.1 kHz.

I may accept there may be differences, what I doubt is that they ARE noticeable (and if noticeable, if they are enough to pick one over the other as "better"). People waste too much time nowadays over these "picky" questions. If it sounds good, then it IS good

Again, that is not true. There are still majority of popular plugins and vendors with no oversampling option especially for plugins intended to be used on all tracks within the project where no latency and low CPU load matters. Not everybody is able to pay for $1000+ CPU and we all know that performance optimisation of DAWs (except Reaper) is still a pain in the ass so every CPU cycle is welcome.
In this scenario I perceive 48kHz as excellent option against 44.1kHz which does not eat much more CPU resources and the sonic result is very close to 88.2 or 96kHz options

[metamorphosis]

This again.

... Okay, so, as anyone who's hung out on these forums long enough knows,
1. 44khz is fine for playback. Very few people can tell the difference between that and 96khz, and usually the difference is caused by the DAC/ADAC working better at 96khz than at 44khz. There are some studies that show subconscious results for inaudible frequencies, both subsonic and hypersonic, but they're debated widely.
2. 96khz is best for recording and processing. The reasons for this include:
a. Most cheap adac/dacs operate with less problems in terms of the rolling off of upper frequencies, at 96khz.
b. When the sample rate is at-or-above 60khz, most decent plugins can push the quantisation noise into the inaudible regions above nyquist.

So, record at 24/96, process at 24/96, dither/resample down to 16/44 for the end result. That's the ideal, but you're CPU-bound of course, depending on your number of tracks etcetera...

More decent info here: https://www.soundonsound.com/sound-advi ... -recording

[fmr]

In this scenario I perceive 48kHz as excellent option against 44.1kHz .../... and the sonic result is very close to 88.2 or 96kHz options

WHAT?

Now I am convinced that it's just a placebo effect.

[MogwaiBoy]

Sonic difference between 48kHz render and downsampled 44.1kHz one is non-audible. Difference between 48kHz original render and 44.1kHz original render is more obvious

48k render then downsampled to 44k still has the benefits of the processing the render at 48k, which are... higher nyquist = less EQ warping + higher automatic low passing on plugins that do it + more space for harmonics to breathe = lower aliasing. Plus instrument and drum machine VSTs will also make use of the higher samplerate for the same reasons. Even when "chopped off" (downsampled to 44.1k) most of these benefits are still there. The difference will be in the high frequencies and manifest as perception of air, depth and clarity. I've tested this myself and the sonic difference is objectively audible - whether it is subtle or major is subjective.

[poonna]

Sonic difference between 48kHz render and downsampled 44.1kHz one is non-audible. Difference between 48kHz original render and 44.1kHz original render is more obvious

48k render then downsampled to 44k still has the benefits of the processing the render at 48k, which are... higher nyquist = less EQ warping + higher automatic low passing on plugins that do it + more space for harmonics to breathe = lower aliasing. Plus instrument and drum machine VSTs will also make use of the higher samplerate for the same reasons. Even when "chopped off" (downsampled to 44.1k) most of these benefits are still there. The difference will be in the high frequencies and manifest as perception of air, depth and clarity. I've tested this myself and the sonic difference is objectively audible - whether it is subtle or major is subjective.

I think I can support this argument. I've tested some plugins that have oversampling, which would require a low-pass filter before downsampling back to the original sample rate. Some of the plugins have a cutoff frequency quite a bit below Nyquist frequency, which for 44.1 kHz, it could start rolling off at 18-19 kHz or earlier. For 48 kHz, the cutoff is pushed a bit further to 20-21 kHz region. Whether this is audible to anyone might be debatable, but it could be the reason why sometimes 48 kHz renders sounds clearer.

[MogwaiBoy]

+1 ^^^

Some plugins have automatic low-pass filters well under 20khz and they are often audible because the filter can reach right down near 10khz - even a -0.1db cut at 12khz over several dozen plugin instances.. is noticeable. This is not to mention phase.

Just having that extra few thousand hertz of frequency info up the top does make a difference. 48khz is awesome

[poshook]

In this scenario I perceive 48kHz as excellent option against 44.1kHz .../... and the sonic result is very close to 88.2 or 96kHz options

WHAT?

Now I am convinced that it's just a placebo effect.

Try mixing whole project at 48k and hear. Do not judge withou trying. You do not need to thank me for great advice.

[Musicisbest]

i wonder as i always thought 44.1 kHZ and 24bit is enough but more and more say especially for sound design, making music/sounds for games etc. that 48 kHZ is a must.

1. is it true?
2. how much impact on the performance (PC/MAC) does it have?

Yes, its true.

[fmr]

Sonic difference between 48kHz render and downsampled 44.1kHz one is non-audible. Difference between 48kHz original render and 44.1kHz original render is more obvious

48k render then downsampled to 44k still has the benefits of the processing the render at 48k, which are... higher nyquist = less EQ warping + higher automatic low passing on plugins that do it + more space for harmonics to breathe = lower aliasing.

Please stop talking nonsense. Dou you EVEN know what alisasing is?

THERE IS NO ALIASING NOW - DAWs HAVE ANTI-ALIASING FILTERS. PLUG-INS HAVE ANTI-ALIASING FILTERS. THE "LOW-PASS" YOU MENTION "IS" THE ANTI-ALIASING FILTER.

What DAW do you use BTW? And do you have measures supporting your statements, or is it "just a feeling"?

Here, take a look of what is an anti-aliasing filter, what is oversampling, and stop talking bullsh!t: https://www.analog.com/en/analog-dialog ... rters.html

https://community.plm.automation.siemen ... a-p/367750

[fmr]

+1 ^^^

Some plugins have automatic low-pass filters well under 20khz and they are often audible because the filter can reach right down near 10khz - even a -0.1db cut at 12khz over several dozen plugin instances.. is noticeable. This is not to mention phase.

Just having that extra few thousand hertz of frequency info up the top does make a difference. 48khz is awesome

Again, more bullsh!t. “Trustable”, alias-free region of the spectrum is from zero Hz to 80% of the bandwidth. This alias-free range is called the frequency span. THIS IS THE RANGE THAT IS NOT AFFECTED BY THE ANTI-ALIAS FILTER. So, at 44.1 kHz, the non-filtered range would be up to around 18-19 kHz, since the Nyquist frequency is 22.05 kHz. No anti-alias filter will act "well under 20khz". This is rubbsih.

If you are experiencing frequency loss at 1-2 kHz (which I doubt) you may have faulty equipment, badly programmed software or simply are doing things wrong. How did you measure this anyway? Or is it just another "feeling"?

I don't dispute that a sample rate of 48 kHz is beter than 44.1 kHz. This is objective. What I dispute is that it is MUCH BETTER. It is only MARGINALLY BETTER. All the rest is placebo effect in your brains. But if you feel better, and work better at 48 kHz, by all means...

This reminds me of all those discussions about which DAW sounds better. I bet you also have one of those DAWs

[Ploki]

Last time i talked about this someone claimed 48k was "music production standard", anyone heard that before?

streaming services are mostly 44,1k
youtube prefers 44,1k
our radio stations are 48k.. (ugh.)

and i do not understand the logic that resampling is a good thing.


i guess the best new DAW feature will be ability to natively record sources at both samples rates, and exporting them both without resampling