How big is the difference in the CPU between 44 and 192 kHz?
- KVRAF
- 9476 posts since 17 Sep, 2002 from Gothenburg Sweden
-
- KVRian
- 926 posts since 14 Dec, 2014
It is going to quadruple CPU plus also RAM, disk space and bandwidth usage, the last one is pretty much guaranteed to cause problems without a SSD hard drive.
So you have to quadruple the specs not only of the CPU, but everything in your computer (except maybe GPU), especially data storage, not only for recorded audio, but freezing is the most common way of dealing with high CPU, so data storage is probably going to need even more than 4x.
I also expect other people (companies' support, collaborators, friends) will need 4x patience to deal with you too.
So you have to quadruple the specs not only of the CPU, but everything in your computer (except maybe GPU), especially data storage, not only for recorded audio, but freezing is the most common way of dealing with high CPU, so data storage is probably going to need even more than 4x.
I also expect other people (companies' support, collaborators, friends) will need 4x patience to deal with you too.
- Banned
- 7624 posts since 13 Nov, 2015 from Norway
-
- addled muppet weed
- 107314 posts since 26 Jan, 2003 from through the looking glass
- KVRAF
- 7521 posts since 9 Jan, 2003 from Saint Louis MO
Bats can hear 113kHz, so clearly 192kHz is just not enough. Gotta go for 256.
Orcas can detect 120kHz, but their sensitivity really drops off around 42kHz, and also they're kind of jerks anyway.
Orcas can detect 120kHz, but their sensitivity really drops off around 42kHz, and also they're kind of jerks anyway.
-
- KVRAF
- 1525 posts since 29 Oct, 2015 from Jupiter 8
high sample rates aren't about improved playback quality though as 16bit/44kHz should indeed already be more than enough for perfect reproductions to the human ear, otherwise plugins wouldn't oversample themself often to at least 88kHz, if not more and some give even options for up to 5 or 10MHz when it comes to processing the audio.
And all these up- and downsample processes have their negative impacts too. At least theoretically, as in practice the oversampling seem to far outweigh the negatives.
Not saying that high sample rates without the need for oversampling are the way to go or will significantly improve your recordings, but they definitely have their plusses over oversampling, much higher overall system resource hungriness aside.
But if i understand correctly, if you are mixing true native high sample rates you are dealing with other problems, as the stuff way above 22kHz is actually still cluttering the bandwidth and can / will cause issues you don't have with just the content humans could hear at standard 44.1k, so low pass filtering above human perception would need to be applied on everything anyway?
And all these up- and downsample processes have their negative impacts too. At least theoretically, as in practice the oversampling seem to far outweigh the negatives.
Not saying that high sample rates without the need for oversampling are the way to go or will significantly improve your recordings, but they definitely have their plusses over oversampling, much higher overall system resource hungriness aside.
But if i understand correctly, if you are mixing true native high sample rates you are dealing with other problems, as the stuff way above 22kHz is actually still cluttering the bandwidth and can / will cause issues you don't have with just the content humans could hear at standard 44.1k, so low pass filtering above human perception would need to be applied on everything anyway?
The GAS is always greener on the other side!
- KVRer
- 29 posts since 12 Feb, 2021
The main reason the average person might need to record at sample rates above 48kHz is if they’re planning on doing sound design that relies on keeping that high frequency content after pitching down/slowing down the playback. Big studios might use higher sample rates simply because they can afford the extra storage space and it makes clients happier.
The most important thing to keep in mind is that even if your interface can support higher sample rates, it’s not necessarily true that your microphones can support them or are flat in that higher frequency range at all. Most engineers choose to use 48kHz for the minor benefits you see in allowing a less steep curve when cutting off all frequency content above the nyquist (exactly half the sample rate) when recording and in plugins that do oversampling, but even the differences here are pretty minimal.
Most decent entry level interfaces can handle 192kHz recording, so please don’t purchase one purely because it can handle higher sample rates, instead pay attention to the much more important features such as the self-noise, number of ins/outs, and the preamps.
The most important thing to keep in mind is that even if your interface can support higher sample rates, it’s not necessarily true that your microphones can support them or are flat in that higher frequency range at all. Most engineers choose to use 48kHz for the minor benefits you see in allowing a less steep curve when cutting off all frequency content above the nyquist (exactly half the sample rate) when recording and in plugins that do oversampling, but even the differences here are pretty minimal.
Most decent entry level interfaces can handle 192kHz recording, so please don’t purchase one purely because it can handle higher sample rates, instead pay attention to the much more important features such as the self-noise, number of ins/outs, and the preamps.
- KVRAF
- 7521 posts since 9 Jan, 2003 from Saint Louis MO
If you use ADAT expander(s), you also sacrifice channels if you run at a higher clock rate. The interfaces I've looked at have one lightpipe in and out at most, and some expanders won't sync at higher than 48kHz anyway.
-
- KVRAF
- 6493 posts since 22 Jan, 2005 from Sweden
I wonder how many plugins are certified for 192 kHz anyway, and hardware even less.
When I bought into Waves Gold 2013, still 3 plugins were not supporting 96kHz even. Came in 2014.
Some seem to think it is just a number this 192.
If to do any jump, go to 88k or 96k.
But I would rather go with Cakewalk/Sonar and use the upsampling feature for selected plugins like synths in particular where you will diminish aliasing. So every plugin can be set for this or not. It will make host feed plugin will 96k clock then plugin will downsample again before entering project. Then plenty headroom in computer to do things that matter on some plugins.
There are some upsampling plugins as well.
When I bought into Waves Gold 2013, still 3 plugins were not supporting 96kHz even. Came in 2014.
Some seem to think it is just a number this 192.
If to do any jump, go to 88k or 96k.
But I would rather go with Cakewalk/Sonar and use the upsampling feature for selected plugins like synths in particular where you will diminish aliasing. So every plugin can be set for this or not. It will make host feed plugin will 96k clock then plugin will downsample again before entering project. Then plenty headroom in computer to do things that matter on some plugins.
There are some upsampling plugins as well.
- KVRAF
- 10562 posts since 3 Feb, 2003 from Finland, Espoo
I highly recommend just sticking with 48kHz (it has become the defacto standard I'd say) and then using upsampling within the plugins that require it (anything that causes harmonic distortion).
The only caveat is that if you run some sample content that is native 96kHz (apparently some rare 3rd party Kontakt sets.. and some stuff from Acustica Audio) then you may perhaps benefit a tiny bit, audio quality wise, from using 96kHz as your native sampling rate.
192kHz is useless in my opinion. If you want to heavily oversample some stuff, just get DDMF Metaplugin for that.
The only caveat is that if you run some sample content that is native 96kHz (apparently some rare 3rd party Kontakt sets.. and some stuff from Acustica Audio) then you may perhaps benefit a tiny bit, audio quality wise, from using 96kHz as your native sampling rate.
192kHz is useless in my opinion. If you want to heavily oversample some stuff, just get DDMF Metaplugin for that.
"Wisdom is wisdom, regardless of the idiot who said it." -an idiot
- KVRist
- 294 posts since 24 Aug, 2017
Nyquist theorem makes that in half, so 192 is really “just” 88,2 in audible range. The real benefit, however, is that your audio processing happens in the inaudible spectrum which means you avoid aliasing/foldback of noise into the audible range. That trade off isn’t worth it to me personally, and 96k is probably more than enough already.
-
- KVRAF
- 1863 posts since 11 Apr, 2008
If you don't know/can't hear benefits of higher sample rate then stay with 44.1kHz for music, 48kHz for video.
You can use oversampling in plugins but it will add additional latency to channel with such effect because it has to upsample/process/downsample. So not recommended for the channels that you want to use to record/play live.
Using oversampling in plugins also cause higher CPU usage. It's not like only higher sample rate in project do that (which some may get such impression by reading many of comments on internet about OS in plugin vs higher sample rate project).
Personally I wouldn't recommend higher sample rate projects or using OS in every single plugin which has this possibility, on older computers like i5/i7. Newer machines Intel i9 or AMD 39**/59** can work in higher sample rates up to 96kHz no problem.
You can use oversampling in plugins but it will add additional latency to channel with such effect because it has to upsample/process/downsample. So not recommended for the channels that you want to use to record/play live.
Using oversampling in plugins also cause higher CPU usage. It's not like only higher sample rate in project do that (which some may get such impression by reading many of comments on internet about OS in plugin vs higher sample rate project).
Personally I wouldn't recommend higher sample rate projects or using OS in every single plugin which has this possibility, on older computers like i5/i7. Newer machines Intel i9 or AMD 39**/59** can work in higher sample rates up to 96kHz no problem.