The Science of Sample Rates *Reading time*

[pjeudy]

The Science of Sample Rates (When Higher Is Better — And When It Isn’t)
http://www.trustmeimascientist.com/2013 ... n-it-isnt/

[anko]

just took the time to read this... and kinda wish at least some others would. nice link pjeudy, to a good summary of the currently available facts about sampling rates in audio.

[Ahornberg]

I can hear the difference between .mp3 and .wav but I can't hear any difference between 44.1kHz and any samplerate above. So I stay on 44.1

[anko]

I can hear the difference between .mp3 and .wav but I can't hear any difference between 44.1kHz and any samplerate above. So I stay on 44.1

yep. the linked article is only about pure sampling rate. everyone agrees mp3 compression damages the sound especially at lower bit rates. i just think a lot of people might not realise that super high sampling rates like 192kHz can actually make sound reproduction accuracy go down.

[ScuzzyEye]

i just think a lot of people might not realise that super high sampling rates like 192kHz can actually make sound reproduction accuracy go down.

To be fair, 192 kHz doesn't make the digital representation of the signal less accurate. Also it's not more accurate. It simply contains higher frequency content. But that content can mess with some amplifiers and speakers, so you end up with worse sounding playback than if those higher frequencies were not there (i.e. you used a lower sampling rate).

[Ahornberg]

i just think a lot of people might not realise that super high sampling rates like 192kHz can actually make sound reproduction accuracy go down.

To be fair, 192 kHz doesn't make the digital representation of the signal less accurate. Also it's not more accurate. It simply contains higher frequency content. But that content can mess with some amplifiers and speakers, so you end up with worse sounding playback than if those higher frequencies were not there (i.e. you used a lower sampling rate).

I think you got it to the point.

[anko]

an increase in timing errors at those super high rates makes distortion go up, which makes reproduction accuracy go down.

but i'm not saying it personally! a number of the very top experts are in the linked article. give it a read.

[ScuzzyEye]

an increase in timing errors at those super high rates makes distortion go up, which makes reproduction accuracy go down.

but i'm not saying it personally! a number of the very top experts are in the linked article. give it a read.

I've read the article. Maybe I should have said, "to be fair to Nyquist, Shannon, Whittaker, and Kotelnikov..." The problem isn't with the math, it's with meat space. The distortions aren't being introduced because the samples are closer together, they're introduced because the electronics involved can't cycle that quickly with accuracy.

So there's nothing wrong with running your soft-synth at 192 kHz, during export. It'll accurately reproduce all the in-audible harmonics of that saw very well. Or it will keep the distortion effect being applied in the box to a recorded guitar from aliasing (again by allowing the in-audible harmonics to be reproduced accurately). But it's best to re-sample it down to 44.1 or 48 kHz before playing it back for the sake of the limitations of the physical world.

[anko]

I've read the article. Maybe I should have said, "to be fair to Nyquist, Shannon, Whittaker, and Kotelnikov..." The problem isn't with the math, it's with meat space. The distortions aren't being introduced because the samples are closer together, they're introduced because the electronics involved can't cycle that quickly with accuracy.

So there's nothing wrong with running your soft-synth at 192 kHz, during export. It'll accurately reproduce all the in-audible harmonics of that saw very well. Or it will keep the distortion effect being applied in the box to a recorded guitar from aliasing (again by allowing the in-audible harmonics to be reproduced accurately). But it's best to re-sample it down to 44.1 or 48 kHz before playing it back for the sake of the limitations of the physical world.

that's correct, except that the problem isn't only in meat space, but also, as the article points out and as you noted above, in any currently available electronics. they simply can't sample at that frequency with great enough timing accuracy, so in both sampling and reproduction distortion is introduced.

my "give it a read" wasn't aimed at you, just a recommendation to anyone who wanted to add to the topic.

[theshoemaker]

So should I than conclude the converter in Reason isn't a good one? I'm no professional at all, I even don't have monitors or a proper sound room yet. But on my Denon in "Pure Direct" mode, where all unnecessary circuitry is switched of I can definitely hear a difference when exporting to 44,1 16 bit dithered or 92 24bit undithered.
Any ideas on that one?

[Ahornberg]

So should I than conclude the converter in Reason isn't a good one? I'm no professional at all, I even don't have monitors or a proper sound room yet. But on my Denon in "Pure Direct" mode, where all unnecessary circuitry is switched of I can definitely hear a difference when exporting to 44,1 16 bit dithered or 92 24bit undithered.
Any ideas on that one?

Do you hear a difference if both sampling frequencies run at the same bit-depth?

[avasopht]

So should I than conclude the converter in Reason isn't a good one? I'm no professional at all, I even don't have monitors or a proper sound room yet. But on my Denon in "Pure Direct" mode, where all unnecessary circuitry is switched of I can definitely hear a difference when exporting to 44,1 16 bit dithered or 92 24bit undithered.
Any ideas on that one?

When running at different samplerates devices may exhibit different behavior, resulting in a different sound.

To confirm that the actual samplerate will make no discernible difference, bounce down at 96khz, import that bounce into a project and then export at 44.1khz.

[theshoemaker]

Okay... Good point. Will give it a try.