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!
HeresJohnny312
Posts: 45
Joined: 26 Feb 2020

29 Feb 2020

guitfnky wrote:
28 Feb 2020
now we’re using tricks to identify the difference. no average listener would ever slow down a piece of audio in order to try to find some “a-ha! see, there is a difference!” moment. of course if you slow things down enough, you’ll hear a difference. because of course, there *is* a difference. the only relevant question from a consumer standpoint is, can you *hear* the difference. the answer for the average consumer is absolutely no. I doubt any one of us on this forum could tell the difference either, under proper scientific testing circumstances (yes, I’ll say it again, a blind test).

the point of the video is that as a practical matter, you’re better off saving your CPU power and drive space by recording at 48k and using oversampling plugins than you are by just recording everything at 96 or 192k. the audible difference is beyond negligible, to the extent that it’s essentially useless for human applications.

now if you’re going to do a lot of mangling where you drastically stretch your audio, sure, in that case it could be useful. but how many of us do that regularly? (remember, key word being ‘drastically’)

Timmy restated the main takeaways of the video very well.
Im sure now that you do not do much processing to your audio and simple productions are fine for non professionals but in the field of Hollywood effects , scoring, foley, and also modern complex electronic music, any professional will agree that there is a lot of pitch modulation (not live music and bands but modern genres) that will benefit from editing and processing in the the highest quality possible. You may think you have not heard it but every transformer, dinosaur, or monster in high budget films were pitch modulated and there is a reason they sound good, it is the higher sample rate. So if the majority of the entertainment industry and modern music production scene isn't practical to you then thats is your opinion and not of those in the not the prevailing entertainment industry. It is not useless in many professional applications.

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

29 Feb 2020

HeresJohnny312 wrote:
29 Feb 2020
guitfnky wrote:
28 Feb 2020
now we’re using tricks to identify the difference. no average listener would ever slow down a piece of audio in order to try to find some “a-ha! see, there is a difference!” moment. of course if you slow things down enough, you’ll hear a difference. because of course, there *is* a difference. the only relevant question from a consumer standpoint is, can you *hear* the difference. the answer for the average consumer is absolutely no. I doubt any one of us on this forum could tell the difference either, under proper scientific testing circumstances (yes, I’ll say it again, a blind test).

the point of the video is that as a practical matter, you’re better off saving your CPU power and drive space by recording at 48k and using oversampling plugins than you are by just recording everything at 96 or 192k. the audible difference is beyond negligible, to the extent that it’s essentially useless for human applications.

now if you’re going to do a lot of mangling where you drastically stretch your audio, sure, in that case it could be useful. but how many of us do that regularly? (remember, key word being ‘drastically’)

Timmy restated the main takeaways of the video very well.
Im sure now that you do not do much processing to your audio and simple productions are fine for non professionals but in the field of Hollywood effects , scoring, foley, and also modern complex electronic music, any professional will agree that there is a lot of pitch modulation (not live music and bands but modern genres) that will benefit from editing and processing in the the highest quality possible. You may think you have not heard it but every transformer, dinosaur, or monster in high budget films were pitch modulated and there is a reason they sound good, it is the higher sample rate. So if the majority of the entertainment industry and modern music production scene isn't practical to you then thats is your opinion and not of those in the not the prevailing entertainment industry. It is not useless in many professional applications.
haha, you’ve clearly never listened to any of my music. :lol:
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

29 Feb 2020

it’s just funny to me that there’s such fervor in people’s defense of doing something so utterly unnecessary, yet to a man, no one is willing to actually do a blind test.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
jam-s
Posts: 3035
Joined: 17 Apr 2015
Location: Aachen, Germany
Contact:

01 Mar 2020

guitfnky wrote:
29 Feb 2020
it’s just funny to me that there’s such fervor in people’s defense of doing something so utterly unnecessary, yet to a man, no one is willing to actually do a blind test.
It's the same with people defending their religious beliefs or homoeopathy or tech cult (apple anyone?).

User avatar
xboix
Posts: 281
Joined: 22 Oct 2019

01 Mar 2020

I personally cannot hear any difference beyond 44.1K so that's what I choose to work at. If you can hear a difference then by all means work at higher sample rates.

Endlessly arguing about it on a forum is kinda pointless, because no-one is going to change their mind!

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

01 Mar 2020

guitfnky wrote:
29 Feb 2020
it’s just funny to me that there’s such fervor in people’s defense of doing something so utterly unnecessary, yet to a man, no one is willing to actually do a blind test.
Here's why the blind test is unnecessary. This clearly shows how some instruments can sound different on different sample rates. Now this is an extreme example, I'll give you that, because it seems Ochen Ks stuff showcases this A LOT more clearly compared to most instruments. But the point is still there. Not always, but SOMETIMES some things sound different depending on the sample rate choice.
This is Chip64 playing, but I remember it was the same issue with his 4mer synth when I trialed it a long time ago.

Now please can we end this discussion. The sample rate choice CAN (in some instances) make a difference.




User avatar
eXode
Posts: 838
Joined: 11 Feb 2015

01 Mar 2020

Very interesting video. It got me thinking, why do they even put 8 times oversampling in their distortion effect? Following the information in the video, putting an internal rate of i.e. 48, or let's say 56 kHz, and then applying a sinc filter on frequencies over 50 kHz should suffice.

I mean, the problem with aliasing, is frequencies above sample rate that reflect down into the audible spectrum, so increasing the sample rate just a bit for some headroom and then cutting everything above out seems that it should solve most of the problems, or am I missing something?

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

01 Mar 2020

eXode wrote:
01 Mar 2020
Very interesting video. It got me thinking, why do they even put 8 times oversampling in their distortion effect? Following the information in the video, putting an internal rate of i.e. 48, or let's say 56 kHz, and then applying a sinc filter on frequencies over 50 kHz should suffice.

I mean, the problem with aliasing, is frequencies above sample rate that reflect down into the audible spectrum, so increasing the sample rate just a bit for some headroom and then cutting everything above out seems that it should solve most of the problems, or am I missing something?
First they try to conquer aliasing or mirroring harmonics from the half sample frequency barrier. That can be done by increasing the SR, the higher the better. But then you get your over-20kHz range polluted by these harmonics, which can intermodulate and appear back in the audible range during subsequent processing. So the range >20kHz should be regularly cleaned up. This leads to combined solution of temporary oversampling for some process then LP-filtering and going back to low SR, which is also cheaper than running at high SR all the time.

User avatar
eXode
Posts: 838
Joined: 11 Feb 2015

01 Mar 2020

orthodox wrote:
01 Mar 2020
eXode wrote:
01 Mar 2020
Very interesting video. It got me thinking, why do they even put 8 times oversampling in their distortion effect? Following the information in the video, putting an internal rate of i.e. 48, or let's say 56 kHz, and then applying a sinc filter on frequencies over 50 kHz should suffice.

I mean, the problem with aliasing, is frequencies above sample rate that reflect down into the audible spectrum, so increasing the sample rate just a bit for some headroom and then cutting everything above out seems that it should solve most of the problems, or am I missing something?
First they try to conquer aliasing or mirroring harmonics from the half sample frequency barrier. That can be done by increasing the SR, the higher the better. But then you get your over-20kHz range polluted by these harmonics, which can intermodulate and appear back in the audible range during subsequent processing. So the range >20kHz should be regularly cleaned up. This leads to combined solution of temporary oversampling for some process then LP-filtering and going back to low SR, which is also cheaper than running at high SR all the time.
I understand that some oversampling can be beneficial, but isn't 8X overkill if you are going to filter out everything above i.e. 40 kHz anyway? It seems that i.e. 1.5X oversampling would be sufficient, and also more performance friendly. Assuming that your host sample rate is at least CD quality, and not lower than that.

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

01 Mar 2020

eXode wrote:
01 Mar 2020
orthodox wrote:
01 Mar 2020


First they try to conquer aliasing or mirroring harmonics from the half sample frequency barrier. That can be done by increasing the SR, the higher the better. But then you get your over-20kHz range polluted by these harmonics, which can intermodulate and appear back in the audible range during subsequent processing. So the range >20kHz should be regularly cleaned up. This leads to combined solution of temporary oversampling for some process then LP-filtering and going back to low SR, which is also cheaper than running at high SR all the time.
I understand that some oversampling can be beneficial, but isn't 8X overkill if you are going to filter out everything above i.e. 20 kHz anyway? It seems that i.e. 1.5X oversampling would be sufficient, and also more performance friendly. Assuming that your host sample rate is at least CD quality, and not lower than that.
For the regular 1/n^2 harmonics level slope (that of triangle waveform, which is typically formed by saturation FX), the aliasing signal level will decrease:
1.5X oversampling: by -12dB
2X: -19dB
4X: -34dB
8X: -47dB
(x): -40 log10(2x-1)

User avatar
eXode
Posts: 838
Joined: 11 Feb 2015

01 Mar 2020

orthodox wrote:
01 Mar 2020
eXode wrote:
01 Mar 2020


I understand that some oversampling can be beneficial, but isn't 8X overkill if you are going to filter out everything above i.e. 20 kHz anyway? It seems that i.e. 1.5X oversampling would be sufficient, and also more performance friendly. Assuming that your host sample rate is at least CD quality, and not lower than that.
For the regular 1/n^2 harmonics level slope (that of triangle waveform, which is typically formed by saturation FX), the aliasing signal level will decrease:
1.5X oversampling: by -12dB
2X: -19dB
4X: -34dB
8X: -47dB
(x): -40 log10(2x-1)
I understand, but again, why does that matter if you filter it out anyway? :)

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

01 Mar 2020

eXode wrote:
01 Mar 2020
I understand, but again, why does that matter if you filter it out anyway? :)
Because aliased/mirrored harmonics are already in the audible range and you can't distinguish them anymore.

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

01 Mar 2020

eXode wrote:
01 Mar 2020
Very interesting video. It got me thinking, why do they even put 8 times oversampling in their distortion effect? Following the information in the video, putting an internal rate of i.e. 48, or let's say 56 kHz, and then applying a sinc filter on frequencies over 50 kHz should suffice.

I mean, the problem with aliasing, is frequencies above sample rate that reflect down into the audible spectrum, so increasing the sample rate just a bit for some headroom and then cutting everything above out seems that it should solve most of the problems, or am I missing something?
it could be more a marketing tactic rather than something that’s actually providing a real benefit. people generally think bigger numbers=better. if you have two great sounding, comparably-priced amp sims for example, and one has 4x oversampling and the other has 8x, the thinking might be that people would opt for the one with the higher number.

just a guess though.
I write good music for good people

https://slowrobot.bandcamp.com/

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

01 Mar 2020

TritoneAddiction wrote:
01 Mar 2020
guitfnky wrote:
29 Feb 2020
it’s just funny to me that there’s such fervor in people’s defense of doing something so utterly unnecessary, yet to a man, no one is willing to actually do a blind test.
Here's why the blind test is unnecessary. This clearly shows how some instruments can sound different on different sample rates. Now this is an extreme example, I'll give you that, because it seems Ochen Ks stuff showcases this A LOT more clearly compared to most instruments. But the point is still there. Not always, but SOMETIMES some things sound different depending on the sample rate choice.
This is Chip64 playing, but I remember it was the same issue with his 4mer synth when I trialed it a long time ago.

Now please can we end this discussion. The sample rate choice CAN (in some instances) make a difference.



I appreciate the examples. in this case I was able to hear the difference, so thank you.

that said, this goes back to the point I made earlier—this test is really not much different than the ones done in the video, in which you can also clearly hear the difference. these are tests, not real-world pieces of music. what I mean when I say you need to do proper blind testing is that you need to test real-world practical scenarios, i.e. test with real music.

and the video doesn’t dispute the fact that there can be audible differences—what they’re pointing out though is that the need for higher sample rates is reduced/eliminated if you use plugins with oversampling. obviously not all plugins have it, so if your need is very specific, sure there could be a difference, but again, is it audible in a real world mix? if you’re using that specific patch above and it’s in isolation, sure. if it’s in a mix, with other stuff going on, doubtful.

I won’t belabor the point any further—I know nothing that either of us say will sway the other, and that’s okay. it’s been an interesting and lively discussion either way. :)
I write good music for good people

https://slowrobot.bandcamp.com/

chaosroyale
Posts: 728
Joined: 05 Sep 2017

01 Mar 2020

Firstly - that is a great video and explains aliasing and oversampling perfectly. There is so much misinformation on sound engineering, and Dan Worrall is one of the few people on youtube who appeals to DAW users who actually knows what he is talking about..

Some people have misunderstood the point about aliasing and sample rates. Yes, you CAN "hear the difference" using synths or effects at different sample rates when that synth or effect can introduce aliasing - but only if it does not have proper oversampling and filtering internally. This is exactly what the video is demonstrating. You are NOT hearing "the difference between 44.1kHZ and 96kHz" - you are hearing aliasing artifacts in a specific situation.

tl,dr; In any DAW, if all your devices have good oversampling, you can not hear the difference between 44.1, 96 or 192 kHz sample rates.

Which brings me to the question: which native Reason devices -if any- have oversampling? I remember that complex-1 does, but anything else?

Anyone know?

chaosroyale
Posts: 728
Joined: 05 Sep 2017

01 Mar 2020

And a quick point to the person who mentioned using ultrasonic samples slowed down to make sound effects - it's very rare to capture ultrasonic samples in the first place, even for movie sound FX.

The field recordings for Jurassic Park for example were done on DAT, which as you know only has a 48kHz sample rate.

Most of the "slowed-down" sound effects you hear in movies of monsters and so on are recorded in the audible range and dropped only by about an octave or two. There is a loss in high frequency content, naturally, but saturation, EQ, and most importantly layering with other higher sounds can be used if more brightness and bite is needed.

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

01 Mar 2020

guitfnky wrote:
01 Mar 2020
I appreciate the examples. in this case I was able to hear the difference, so thank you.

that said, this goes back to the point I made earlier—this test is really not much different than the ones done in the video, in which you can also clearly hear the difference. these are tests, not real-world pieces of music. what I mean when I say you need to do proper blind testing is that you need to test real-world practical scenarios, i.e. test with real music.

and the video doesn’t dispute the fact that there can be audible differences—what they’re pointing out though is that the need for higher sample rates is reduced/eliminated if you use plugins with oversampling. obviously not all plugins have it, so if your need is very specific, sure there could be a difference, but again, is it audible in a real world mix? if you’re using that specific patch above and it’s in isolation, sure. if it’s in a mix, with other stuff going on, doubtful.

I won’t belabor the point any further—I know nothing that either of us say will sway the other, and that’s okay. it’s been an interesting and lively discussion either way. :)
My first examples in this thread were actually "real-world pieces of music". They were part of this track that I just finished. I just thought it would be easier to isolate one sound.
Pretty happy with how this tune turned out actually.


Anyway you're right. I don't think any of us will change our stance on this topic. I think sample rate matters. You don't. Let's not drag this discussion out any longer. Thanks for taking the time.

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

01 Mar 2020

chaosroyale wrote:
01 Mar 2020
Which brings me to the question: which native Reason devices -if any- have oversampling? I remember that complex-1 does, but anything else?

Anyone know?
I'm sure that every Reason synth starting from Subtractor does antialiasing for its oscillator waveforms. Maybe it does not involve oversampling, though it's a common thing for synths since long time ago.

User avatar
eXode
Posts: 838
Joined: 11 Feb 2015

01 Mar 2020

orthodox wrote:
01 Mar 2020
chaosroyale wrote:
01 Mar 2020
Which brings me to the question: which native Reason devices -if any- have oversampling? I remember that complex-1 does, but anything else?

Anyone know?
I'm sure that every Reason synth starting from Subtractor does antialiasing for its oscillator waveforms. Maybe it does not involve oversampling, though it's a common thing for synths since long time ago.
Subtractor alias a lot, so I doubt it. For Thor I think they used some bandwidth limiting, but it still aliases when you do some modulation (I.e. FM or sync). Even Europa alias, although it's cleaner than some of their older stuff.

The fact is that it seems that many of the Reason devices have implemented very little to counter aliasing, sadly. Which probably is one of the reasons they are so CPU efficient.

chaosroyale
Posts: 728
Joined: 05 Sep 2017

01 Mar 2020

Seems like a perfect chance for RS to upgrade all the instruments - add a 2x oversampling switch on the back! That's gotta appeal to new users looking for "the best" sounds, as well as older farts like me who wish the devices would be upgraded occasionally.
eXode wrote:
01 Mar 2020
orthodox wrote:
01 Mar 2020


I'm sure that every Reason synth starting from Subtractor does antialiasing for its oscillator waveforms. Maybe it does not involve oversampling, though it's a common thing for synths since long time ago.
Subtractor alias a lot, so I doubt it. For Thor I think they used some bandwidth limiting, but it still aliases when you do some modulation (I.e. FM or sync). Even Europa alias, although it's cleaner than some of their older stuff.

The fact is that it seems that many of the Reason devices have implemented very little to counter aliasing, sadly. Which probably is one of the reasons they are so CPU efficient.

User avatar
BRIGGS
Posts: 2135
Joined: 25 Sep 2015
Location: Orange County California

05 Mar 2020

A follow up video:

r11s

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

06 Mar 2020

chaosroyale wrote:
01 Mar 2020
Firstly - that is a great video and explains aliasing and oversampling perfectly. There is so much misinformation on sound engineering, and Dan Worrall is one of the few people on youtube who appeals to DAW users who actually knows what he is talking about..

Some people have misunderstood the point about aliasing and sample rates. Yes, you CAN "hear the difference" using synths or effects at different sample rates when that synth or effect can introduce aliasing - but only if it does not have proper oversampling and filtering internally. This is exactly what the video is demonstrating. You are NOT hearing "the difference between 44.1kHZ and 96kHz" - you are hearing aliasing artifacts in a specific situation.

tl,dr; In any DAW, if all your devices have good oversampling, you can not hear the difference between 44.1, 96 or 192 kHz sample rates.

Which brings me to the question: which native Reason devices -if any- have oversampling? I remember that complex-1 does, but anything else?

Anyone know?
Great points, saved me the time to type the same response.

There is a similar situation with some cheeper converters, where higher sample rates DID sound better because they used crappy filters on the lower sample rates to save $$$. But when I compared high end converters at different rates, it was very difficult to tell if there was ANY difference - if there WAS, I couldn't 100% identify it.

So yes, sample rates CAN make a difference in some cases, but not for the reasons you may think (not because it's "more accurate" to capture more data, or because there are "gaps" in the sound etc.).

The main point being you cannot unequivocally say higher rates are always better, but in most cases you can say they present their own set of issues.
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

06 Mar 2020

I'm going to include ultrasonic signal metering and filtering out in my next RE. So that one would be able to get rid of potentially intermodulating content in the ultrasonic range when using high sample rates.

User avatar
Boombastix
Competition Winner
Posts: 1929
Joined: 18 May 2018
Location: Bay Area, CA

06 Mar 2020

BRIGGS wrote:
05 Mar 2020
A follow up video:

Dan Worrall and his "old fart ears" :o . He's on fire! :lol:
10% off at Waves with link: https://www.waves.com/r/6gh2b0
Disclaimer - I get 10% as well.

User avatar
guitfnky
Posts: 4408
Joined: 19 Jan 2015

06 Mar 2020

:lol: that was great 👍🏼
I write good music for good people

https://slowrobot.bandcamp.com/

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 12 guests