Samplerates: the higher the better, right? (Dan Worrall)

Have an urge to learn, or a calling to teach? Want to share some useful Youtube videos? Do it here!
User avatar
BRIGGS
Posts: 2133
Joined: 25 Sep 2015
Location: Orange County California

27 Feb 2020

:puf_smile: Super interesting:

r11s

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

27 Feb 2020

Love it, great information and examples.
Selig Audio, LLC

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

27 Feb 2020

Nice illustrations.

HeresJohnny312
Posts: 45
Joined: 26 Feb 2020

27 Feb 2020

Higher Sample rates(96-192khz) imo are beneficial in the same token that 60-120 fps video are benficial fro slowing down and keeping as much information as possible for fluid quality even at low pitches. For instance in movie editing, 192khz is standard in high budget productions for effects and creative sounds. You'll realize that alot of dinosaurs and monsters are just pitched animals with high quality sample rates and added effects!

User avatar
demt
Posts: 1357
Joined: 16 Sep 2016
Contact:

27 Feb 2020

crumbs the truth at last its been years n years n years praps this that n definatly
Reason 12 ,gear4 music sdp3 stage piano .nektar gxp 88,behringer umc1800 .line6 spider4 30
hear scince reason 2.5

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

27 Feb 2020

HeresJohnny312 wrote:
27 Feb 2020
Higher Sample rates(96-192khz) imo are beneficial in the same token that 60-120 fps video are benficial fro slowing down and keeping as much information as possible for fluid quality even at low pitches. For instance in movie editing, 192khz is standard in high budget productions for effects and creative sounds. You'll realize that alot of dinosaurs and monsters are just pitched animals with high quality sample rates and added effects!
But note one major difference - not all convertors are designed to capture super sonic frequencies even when set to higher rates. Some simply use more gentle filters, still starting to cut off frequencies above 20 kHz but at a more gentle slope with higher sample rates. This means that little additional content is created with most A/Ds when recording at higher rates.

If you speak to those who record things like bats (or simply google it), you'll find it requires special rigs to ensure you're really recording those super sonic frequencies, because if you don't actually capture them then you won't gain anything from slowing the recordings down later…
Selig Audio, LLC

User avatar
TheDragonborg
Posts: 285
Joined: 10 Feb 2020

27 Feb 2020

I work only on 44Khz at 16 bit nothing else really matters for stereo music as that's plenty for that.

HeresJohnny312
Posts: 45
Joined: 26 Feb 2020

27 Feb 2020

These are the basic benefits in a nutshell. More information captured is better because less is approximated and there is more content recorded, then when dumping to cd quality or otherwise known as 16bit 44.1 there is absolutely the highest possible quality crushed into that 44.1, now in the same token as visual you can never shoot 720 and render in 1080 its still crap. but you can however shoot in 4k and crush to 1080 and it will look better than the 1080 shot and rendered to 1080. It basically ensuring highest quality, higher bandwidth, and also more information as well as higher headroom, now I think those few benefits are small yet worth making your cpu work a bit harder.... just my rant sorry! Most people can't tell but, then again most people can't decipher digital compression from analog optical compression.

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

28 Feb 2020

HeresJohnny312 wrote:
27 Feb 2020
These are the basic benefits in a nutshell. More information captured is better because less is approximated and there is more content recorded, then when dumping to cd quality or otherwise known as 16bit 44.1 there is absolutely the highest possible quality crushed into that 44.1, now in the same token as visual you can never shoot 720 and render in 1080 its still crap. but you can however shoot in 4k and crush to 1080 and it will look better than the 1080 shot and rendered to 1080. It basically ensuring highest quality, higher bandwidth, and also more information as well as higher headroom, now I think those few benefits are small yet worth making your cpu work a bit harder.... just my rant sorry! Most people can't tell but, then again most people can't decipher digital compression from analog optical compression.
You have made what I feel is a false equivalence with the video comparison. Video resolution is entirely dependent on screen size and distance. If you view a video on a phone from 2 feet away vs viewing a projection in a huge theater, you NEED higher resolution for the theater than for the phone. But the audio in BOTH cases needs the same sample rate because frequency response does not depend on the size of the room or the distance from the speakers.

You also mention more data is better, which is contradicted by the video in this thread because in some cases the higher frequency data will cause problems in the audible range. Also, with all sample rates NOTHING is "approximated" - it's either captured or it is not. There is no need to capture super sonic frequencies, just as there is no reason to capture ultra violet light with a video camera (which is a better comparison between audio and video). Or to put it another way, we cannot see ultraviolet light so there is no benefit capturing it, same as we cannot hear super sonic frequencies so there is no value in capturing them.

There IS value in oversampling to prevent aliasing, but that's an entirely different concept.

Side note: if you DO want to compare video resolution to audio, compare it to bit depth. In the theater vs phone example above, one could say that 12-16 bit audio is fine for the phone since it's played at such a low SPL level (so you don't need a wide dynamic range) if just because you are closer to the phone. But in the theater, the SPL could be much higher, and so you would be able to take advantage of a higher dynamic range such as 20-24 bit. Make sense?
Selig Audio, LLC

User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

selig wrote:
28 Feb 2020
HeresJohnny312 wrote:
27 Feb 2020
These are the basic benefits in a nutshell. More information captured is better because less is approximated and there is more content recorded, then when dumping to cd quality or otherwise known as 16bit 44.1 there is absolutely the highest possible quality crushed into that 44.1, now in the same token as visual you can never shoot 720 and render in 1080 its still crap. but you can however shoot in 4k and crush to 1080 and it will look better than the 1080 shot and rendered to 1080. It basically ensuring highest quality, higher bandwidth, and also more information as well as higher headroom, now I think those few benefits are small yet worth making your cpu work a bit harder.... just my rant sorry! Most people can't tell but, then again most people can't decipher digital compression from analog optical compression.
You have made what I feel is a false equivalence with the video comparison. Video resolution is entirely dependent on screen size and distance. If you view a video on a phone from 2 feet away vs viewing a projection in a huge theater, you NEED higher resolution for the theater than for the phone. But the audio in BOTH cases needs the same sample rate because frequency response does not depend on the size of the room or the distance from the speakers.

You also mention more data is better, which is contradicted by the video in this thread because in some cases the higher frequency data will cause problems in the audible range. Also, with all sample rates NOTHING is "approximated" - it's either captured or it is not. There is no need to capture super sonic frequencies, just as there is no reason to capture ultra violet light with a video camera (which is a better comparison between audio and video). Or to put it another way, we cannot see ultraviolet light so there is no benefit capturing it, same as we cannot hear super sonic frequencies so there is no value in capturing them.

There IS value in oversampling to prevent aliasing, but that's an entirely different concept.

Side note: if you DO want to compare video resolution to audio, compare it to bit depth. In the theater vs phone example above, one could say that 12-16 bit audio is fine for the phone since it's played at such a low SPL level (so you don't need a wide dynamic range) if just because you are closer to the phone. But in the theater, the SPL could be much higher, and so you would be able to take advantage of a higher dynamic range such as 20-24 bit. Make sense?
But the samething can be said about hearing frequencies close to Nyquist due to anti-aliasing and sharp filters. Though the perception intensity is a lot harder with audio. Our ability to hear a difference between 44.1 and 48 kHz reproduced waveform then projected through a PA is a lot harder to distinguish than a 1080p to 720p resolution due to maxing out our perception ability in the audio spectrum. We would have to rely on modulation/distrotion in audio or the ability to clearly hear 17-20k to notice a substantial difference. Where as a change of 720p to 108p has a more grandeur change. If you want to use that comparison, I would say think more on the lines between 4k to 5k . . . or a point where its hard to determine our ability to perceive.

But normally I would compare data rates to data rates. SPS <-> FPS as its based on a number of samples, ie. the frequency of what can be recorded. An object or phenomena can't move no faster or slower than what can physically be captured in a digital domain. Lower frame rates will make items look to not move fluid or fast at all (think action scenes) while a lower sample rate will cause a reduction of higher frequencies simply because information is not even being captured.
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

28 Feb 2020

Kalm wrote:
28 Feb 2020
But the samething can be said about hearing frequencies close to Nyquist due to anti-aliasing and sharp filters. Though the perception intensity is a lot harder with audio. Our ability to hear a difference between 44.1 and 48 kHz reproduced waveform then projected through a PA is a lot harder to distinguish than a 1080p to 720p resolution due to maxing out our perception ability in the audio spectrum. We would have to rely on modulation/distrotion in audio or the ability to clearly hear 17-20k to notice a substantial difference. Where as a change of 720p to 108p has a more grandeur change. If you want to use that comparison, I would say think more on the lines between 4k to 5k . . . or a point where its hard to determine our ability to perceive.

But normally I would compare data rates to data rates. SPS <-> FPS as its based on a number of samples, ie. the frequency of what can be recorded. An object or phenomena can't move no faster or slower than what can physically be captured in a digital domain. Lower frame rates will make items look to not move fluid or fast at all (think action scenes) while a lower sample rate will cause a reduction of higher frequencies simply because information is not even being captured.
The 17-20k range is *EQUALLY* well represented by 44.1k and 48k sample rates.
Any comparison with picture resolution is irrelevant, simply because of essentially different mechanisms of visual and aural perception. We see things in space domain and the quality depends on correct SAMPLE representation, which can be improved by increasing the resolution. While the hearing is done directly in frequency domain and we don't hear anything above some frequency limit at all as we don't have receptors for that.

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

28 Feb 2020

never understood the desire to record at such high sample rates. there's a reason the industry still uses 44.1 as its primary sample format.

I defy anyone here who still thinks that they can discern the difference in quality between something recorded at 192k and 48k after watching this video to do some blind testing and prove it to yourselves.


it seems a strange point of pride for some to tax their CPUs 2-4 times more than necessary.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
TritoneAddiction
Competition Winner
Posts: 4219
Joined: 29 Aug 2015
Location: Sweden

28 Feb 2020

I don't care what anyone says. When I'm working on music in Reason I can hear a difference between 44100 and 96000. Sure it's not a night and day difference, but with some sounds there's a difference. If I'm playing my mix in 96000 things sound smoother, less harsh. Drum samples, synths, they can both be affected by the sample rate.
So as long as my computer can chug along in 96000 that's what I'll do.

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

28 Feb 2020

TritoneAddiction wrote:
28 Feb 2020
I don't care what anyone says. When I'm working on music in Reason I can hear a difference between 44100 and 96000. Sure it's not a night and day difference, but with some sounds there's a difference. If I'm playing my mix in 96000 things sound smoother, less harsh. Drum samples, synths, they can both be affected by the sample rate.
So as long as my computer can chug along in 96000 that's what I'll do.
Sure they would sound different, that's what the OP video is about. Yet I think, if you convert the 96k mixdown to 44.1k, it won't lose its qualities.

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

28 Feb 2020

TritoneAddiction wrote:
28 Feb 2020
I don't care what anyone says. When I'm working on music in Reason I can hear a difference between 44100 and 96000. Sure it's not a night and day difference, but with some sounds there's a difference. If I'm playing my mix in 96000 things sound smoother, less harsh. Drum samples, synths, they can both be affected by the sample rate.
So as long as my computer can chug along in 96000 that's what I'll do.
nah.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

guitfnky wrote:
28 Feb 2020
TritoneAddiction wrote:
28 Feb 2020
I don't care what anyone says. When I'm working on music in Reason I can hear a difference between 44100 and 96000. Sure it's not a night and day difference, but with some sounds there's a difference. If I'm playing my mix in 96000 things sound smoother, less harsh. Drum samples, synths, they can both be affected by the sample rate.
So as long as my computer can chug along in 96000 that's what I'll do.
nah.
There is definitely an audible difference depending on the source
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

orthodox wrote:
28 Feb 2020
Kalm wrote:
28 Feb 2020
But the samething can be said about hearing frequencies close to Nyquist due to anti-aliasing and sharp filters. Though the perception intensity is a lot harder with audio. Our ability to hear a difference between 44.1 and 48 kHz reproduced waveform then projected through a PA is a lot harder to distinguish than a 1080p to 720p resolution due to maxing out our perception ability in the audio spectrum. We would have to rely on modulation/distrotion in audio or the ability to clearly hear 17-20k to notice a substantial difference. Where as a change of 720p to 108p has a more grandeur change. If you want to use that comparison, I would say think more on the lines between 4k to 5k . . . or a point where its hard to determine our ability to perceive.

But normally I would compare data rates to data rates. SPS <-> FPS as its based on a number of samples, ie. the frequency of what can be recorded. An object or phenomena can't move no faster or slower than what can physically be captured in a digital domain. Lower frame rates will make items look to not move fluid or fast at all (think action scenes) while a lower sample rate will cause a reduction of higher frequencies simply because information is not even being captured.
The 17-20k range is *EQUALLY* well represented by 44.1k and 48k sample rates.
Any comparison with picture resolution is irrelevant, simply because of essentially different mechanisms of visual and aural perception. We see things in space domain and the quality depends on correct SAMPLE representation, which can be improved by increasing the resolution. While the hearing is done directly in frequency domain and we don't hear anything above some frequency limit at all as we don't have receptors for that.
It's not fully equal if intermodulation is occurring. Our response to reconstruction may differ anytime the quality of aliasing or anti-aliasing happens. But in visual the samething happens. They are BOTH represented by a frequency domain. In this instance it would be color and no amount perception will ever allow you to see beyond a human limited frequency domain (infrared and ultraviolet) unless filtered to our spectrum somehow. But sample best entities would be represented through bits as that is a digital construction over something like frequency which is available in nature. We capture one and distribute in the other. We capture frequency and distribute samples to define a resolution, though resolution is actually occurring in the frequency domain since we as humans have limitations.
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

28 Feb 2020

Kalm wrote:
28 Feb 2020
guitfnky wrote:
28 Feb 2020


nah.
There is definitely an audible difference depending on the source
I defy anyone here who still thinks that they can discern the difference in quality between something recorded at 192k and 48k after watching this video to do some blind testing and prove it to yourselves.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

guitfnky wrote:
28 Feb 2020
Kalm wrote:
28 Feb 2020


There is definitely an audible difference depending on the source
I defy anyone here who still thinks that they can discern the difference in quality between something recorded at 192k and 48k after watching this video to do some blind testing and prove it to yourselves.
I’ve heard it in Reason, certain sounds construct differently simply from changing the sample rate of your converter in your interface. Hence what TritoneAddiction was saying. I can try to create an example but it’s heavily dependent on the source. Theirs videos on YouTube though that show this however
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

28 Feb 2020

Kalm wrote:
28 Feb 2020
guitfnky wrote:
28 Feb 2020


I’ve heard it in Reason, certain sounds construct differently simply from changing the sample rate of your converter in your interface. Hence what TritoneAddiction was saying. I can try to create an example but it’s heavily dependent on the source. Theirs videos on YouTube though that show this however
trying to recreate this is not a blind test.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

guitfnky wrote:
28 Feb 2020
Kalm wrote:
28 Feb 2020


I’ve heard it in Reason, certain sounds construct differently simply from changing the sample rate of your converter in your interface. Hence what TritoneAddiction was saying. I can try to create an example but it’s heavily dependent on the source. Theirs videos on YouTube though that show this however
trying to recreate this is not a blind test.
I already stated its been done on Youtube. Recreation is not a blind test because its factual. There is an audible difference based on the source. Hence if I recreate the change I will hear it. No blind needed
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

28 Feb 2020

Kalm wrote:
28 Feb 2020
guitfnky wrote:
28 Feb 2020


trying to recreate this is not a blind test.
I already stated its been done on Youtube. Recreation is not a blind test because its factual. There is an audible difference based on the source. Hence if I recreate the change I will hear it. No blind needed
sure it’s needed. you keep stating you can definitively hear the difference. watching someone else claim they can hear a difference on youtube is completely irrelevant.

I’m not talking about a sine wave test like in the video. I’m talking about real world mixes. can you tell the difference? I very much doubt it. if you aren’t willing to prove it (to yourself at the very least) by doing a proper blind test, then why would anyone take your claim seriously?
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

28 Feb 2020

Kalm wrote:
28 Feb 2020
orthodox wrote:
28 Feb 2020


The 17-20k range is *EQUALLY* well represented by 44.1k and 48k sample rates.
Any comparison with picture resolution is irrelevant, simply because of essentially different mechanisms of visual and aural perception. We see things in space domain and the quality depends on correct SAMPLE representation, which can be improved by increasing the resolution. While the hearing is done directly in frequency domain and we don't hear anything above some frequency limit at all as we don't have receptors for that.
It's not fully equal if intermodulation is occurring. Our response to reconstruction may differ anytime the quality of aliasing or anti-aliasing happens. But in visual the samething happens. They are BOTH represented by a frequency domain. In this instance it would be color and no amount perception will ever allow you to see beyond a human limited frequency domain (infrared and ultraviolet) unless filtered to our spectrum somehow. But sample best entities would be represented through bits as that is a digital construction over something like frequency which is available in nature. We capture one and distribute in the other. We capture frequency and distribute samples to define a resolution, though resolution is actually occurring in the frequency domain since we as humans have limitations.
Intermodulation can only occur as a result of modulation, ie only when the signal is being processed. That has nothing to do with the quality of representation. If you get aliasing artifacts just from A/D conversion, this can be corrected by an LP filter before the conversion.

Here is an example that shows why image and audio digital representations are totally different. If you downsample a 1080 picture to 720, then upsample them both to 4K, they will be different and look different. Buf if you downsample a 48k audio to 44.1k and then upsample them both to 192k, they will be identical sample-wise, provided the original 48k audio did not contain frequencies above 22k, which we don't hear.

User avatar
TritoneAddiction
Competition Winner
Posts: 4219
Joined: 29 Aug 2015
Location: Sweden

28 Feb 2020

guitfnky wrote:
28 Feb 2020
Kalm wrote:
28 Feb 2020


I already stated its been done on Youtube. Recreation is not a blind test because its factual. There is an audible difference based on the source. Hence if I recreate the change I will hear it. No blind needed
sure it’s needed. you keep stating you can definitively hear the difference. watching someone else claim they can hear a difference on youtube is completely irrelevant.

I’m not talking about a sine wave test like in the video. I’m talking about real world mixes. can you tell the difference? I very much doubt it. if you aren’t willing to prove it (to yourself at the very least) by doing a proper blind test, then why would anyone take your claim seriously?
Here you go.
This is just one sound from a track I'm working on.
Again there's not a BIG difference. But there IS a difference. Now if that little difference is there on every sound in a song, there's gonna be a different "feel" in the overall mix.




User avatar
Kalm
Posts: 554
Joined: 03 Jun 2016
Location: Austin
Contact:

28 Feb 2020

guitfnky wrote:
28 Feb 2020
Kalm wrote:
28 Feb 2020


I already stated its been done on Youtube. Recreation is not a blind test because its factual. There is an audible difference based on the source. Hence if I recreate the change I will hear it. No blind needed
sure it’s needed. you keep stating you can definitively hear the difference. watching someone else claim they can hear a difference on youtube is completely irrelevant.

I’m not talking about a sine wave test like in the video. I’m talking about real world mixes. can you tell the difference? I very much doubt it. if you aren’t willing to prove it (to yourself at the very least) by doing a proper blind test, then why would anyone take your claim seriously?
Cause you're still hearing a source. but do it yourself. I have a Quantum so I can reproduce/hear samples up to 192. Open something like TRK-01 KICK and export the same note at each sample rate you want to test. TRK-01 can playback at different sample rates. the first default kick is a prime example. Export that same sample rate kick out of your DAW. PLAYBACK all the samples at 48 kHz. you will hear an obvious difference. I'm telling you this so you can do the experiment yourself without relying on me doing something since that's just as problematic as a youtube video.

Here are 4 kick samples I created using the same note processed via MIDI through Reason's export. These included changing the sample rate within TRK-01 for the sample then exporting out of Reason at the same sample rate.

https://www.dropbox.com/sh/vxigvp12a9a7 ... ltQRa?dl=0
Courtesy of The Brew | Watch My Tutorials | Mac Mini Intel i7 Quad-Core | 16 GB RAM | Samsung 850 EVO 250 GB | Reason 11 Suite | Studio One 5 Professional | Presonus Quantum | Komplete Kontrol 49 MK2 | Event Opals | Follow me on Instagram

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 11 guests