Reply to this topicStart new topic
> Audio Aliasing Explained
Mertay
post Aug 30 2015, 02:25 PM
Post #1


GMC:er
Group Icon

Group: GMC Senior
Posts: 2.598
Joined: 27-May 13
From: Turkey / izmir
Member No.: 18.294





For the advanced users...


--------------------
Go to the top of the page
 
+Quote Post
fzalfa
post Aug 31 2015, 09:52 AM
Post #2


Learning Roadie
*

Group: Members
Posts: 1.662
Joined: 10-March 15
From: France, provence, vaucluse, carpentras
Member No.: 20.796



interrsting !

Laurent


--------------------

Yes, i love badges.....
Go to the top of the page
 
+Quote Post
Mertay
post Aug 31 2015, 10:40 AM
Post #3


GMC:er
Group Icon

Group: GMC Senior
Posts: 2.598
Joined: 27-May 13
From: Turkey / izmir
Member No.: 18.294



There are a lot of if's/but's...when it comes to theory but this video explains the situation nicely. I shared cause as someone gets into sound tech. more he'll start to notice mentioning on aliasing.

This is one of the reasons why I suggest working at 96khz (and not consider upsampling plug-ins) and trying to avoid too many plug-ins but getting the sound from analog (like at least using analog pedals if one can't mic. his/her amp in his house) as much as possible.

When I was a student, me and my friends recorded some stuff in different sample rates and made listening tests (including converting them back to 44.1khz) and higher sample rates had noticeably better depth feeling. Some will say humans can't hear those high freq.s but as shown in video its not that simple.


--------------------
Go to the top of the page
 
+Quote Post
Rammikin
post Aug 31 2015, 05:23 PM
Post #4


Experienced Rock Star
*

Group: Members
Posts: 731
Joined: 4-November 10
Member No.: 11.529



As you say, it's not that simple.

First, any quality distortion plugin, including probably every commercial amp simulator plugin ever made, has built-in oversampling. In other words, in practice, it's unusual to encounter aliasing problems. There are many reasons why one might want to use outboard analog gear, but avoiding aliasing generally isn't one of them.

Second, running a session at 96KHz means, generally, using twice as much cpu on your computer than running at 48KHz (and your recordings will consume twice the disk space), so there's a significant cost associated with working at 96KHz.


--------------------
Go to the top of the page
 
+Quote Post
Mertay
post Aug 31 2015, 06:52 PM
Post #5


GMC:er
Group Icon

Group: GMC Senior
Posts: 2.598
Joined: 27-May 13
From: Turkey / izmir
Member No.: 18.294



True, though in a mix we must consider tape/tube etc. emulating plug-ins as well and also we usually don't know whats going on with freeware stuff. It's not worth getting stressed for home guys, but avoiding it is just simple enough.

As the cost of working with 96khz, that really depends on musical goals not the computer unless its really old or has a bad system (mine is 6 years old). Its the reason I recommended pedals-connect or mic.ing amps, better sound and more cost effective than renewing the computer (and the guitar feels great! smile.gif )

Also its said that (couldn't personally test this) our "consumer grade" soundcard converters work way better at higher sample rates cause on higher sample-rates they do less aggressive filtering.

This post has been edited by Mertay: Aug 31 2015, 07:03 PM


--------------------
Go to the top of the page
 
+Quote Post
Todd Simpson
post Sep 3 2015, 01:07 AM
Post #6


GMC:er
Group Icon

Group: GMC Instructor
Posts: 14.138
Joined: 23-December 09
From: Atlanta, Georgia, USA
Member No.: 8.794



Cool vid smile.gif There is certainly a healthy debate to be had about how high to set your sampling rate. There are pretty good arguments both for and against. Like many things, it often comes down to the person/persons doing the session, and the gear on hand. If you are working with an older laptop and a nearly full hard drive, 96khz can be very unattractive as an option. If you have plenty of cpu headroom and wads of disk space, it's far less of issue in many cases.

For me the sad part always comes when trying to downsample to common distribution formats like 44.1k mp3 files. This too is an often bemoaned idea. Starting with something that sounds great and then stepping on it so that it works with folks phones. It's worth considering in any case what the best options are given the variables so well done sharing the vid smile.gif


QUOTE (Mertay @ Aug 31 2015, 01:52 PM) *
True, though in a mix we must consider tape/tube etc. emulating plug-ins as well and also we usually don't know whats going on with freeware stuff. It's not worth getting stressed for home guys, but avoiding it is just simple enough.

As the cost of working with 96khz, that really depends on musical goals not the computer unless its really old or has a bad system (mine is 6 years old). Its the reason I recommended pedals-connect or mic.ing amps, better sound and more cost effective than renewing the computer (and the guitar feels great! smile.gif )

Also its said that (couldn't personally test this) our "consumer grade" soundcard converters work way better at higher sample rates cause on higher sample-rates they do less aggressive filtering.



--------------------
Go to the top of the page
 
+Quote Post
Mertay
post Sep 3 2015, 01:32 AM
Post #7


GMC:er
Group Icon

Group: GMC Senior
Posts: 2.598
Joined: 27-May 13
From: Turkey / izmir
Member No.: 18.294



QUOTE (Todd Simpson @ Sep 3 2015, 12:07 AM) *
For me the sad part always comes when trying to downsample to common distribution formats like 44.1k mp3 files. This too is an often bemoaned idea. Starting with something that sounds great and then stepping on it so that it works with folks phones. It's worth considering in any case what the best options are given the variables so well done sharing the vid smile.gif


I have some nice links for you on downsampling smile.gif

http://src.infinitewave.ca/

I worked a lot with this program at the time, even the free version is awesome. It uses a different math than DAW's use so it eliminates mathematical debates like work an convert 44.100 vs work 48.000 to convert 44.100...been advising 48.000 (or higher) ever since;

http://www.voxengo.com/product/r8brain/

This post has been edited by Mertay: Sep 3 2015, 01:35 AM


--------------------
Go to the top of the page
 
+Quote Post
Rammikin
post Sep 3 2015, 05:52 AM
Post #8


Experienced Rock Star
*

Group: Members
Posts: 731
Joined: 4-November 10
Member No.: 11.529



The best advice is to use the sample rate you'll eventually be delivering. In other words, the best downsampling is no downsampling smile.gif. That's the only mathematical issue regarding sample rates for which there is no debate. So if your goal is to produce a CD, you should use 44.1KHz.

If, after considering the pros and cons, you want to work with a higher sample rate than that, then use an even multiple. For example, if your goal is to deliver something at 44.1KHz, but you want to work at a higher sample rate, then use 88.2KHz. That will minimize the downsampling artifacts.

However, anyone who is that concerned about aliasing might want to consider recording to tape instead of a computer smile.gif.


--------------------
Go to the top of the page
 
+Quote Post
Mertay
post Sep 3 2015, 03:22 PM
Post #9


GMC:er
Group Icon

Group: GMC Senior
Posts: 2.598
Joined: 27-May 13
From: Turkey / izmir
Member No.: 18.294



QUOTE (Rammikin @ Sep 3 2015, 04:52 AM) *
That's the only mathematical issue regarding sample rates for which there is no debate. So if your goal is to produce a CD, you should use 44.1KHz...


Mathematical issue is true for realtime src (we can even include hardware for this) but not true for non-realtime src. Its the reason I shared the voxengo freeware, works non-realtime and uses different algorithm. There are other payware products like izotope rx, weiss saracon...many mastering engineers prefer these over the DAW's they use for src for such reason.

I will add though there are still artifacts on non-realtime sampling, but its mostly (!) related to nyqist point differences (basically the high-cut filter's effect we can hear when doing a phase inversion test between a processed and non-processed sample, filters can be seen on passband examples on src website).

The debate here is that (lets say for a GMC member doing his first album at home) though this artifact occurs on src, when recording our converters use less aggressive filtering (specially the affordable ones smile.gif ) which a reason why we can get more depth feel when recording higher sample rates then down sample to cd format. This is really a personal thing, if one can't hear extra depth I can't argue but I did make this test (wasn't easy) best possible with my friends when we were students and it was a rewarding experience.

Sure things still are debatable and I respect that of course, like on higher sample rates inter-modulation-distortion can happen etc. but its always best to invite interested members to make their own tests if they like pro-audio.

This post has been edited by Mertay: Sep 3 2015, 03:28 PM


--------------------
Go to the top of the page
 
+Quote Post
Todd Simpson
post Sep 5 2015, 12:41 AM
Post #10


GMC:er
Group Icon

Group: GMC Instructor
Posts: 14.138
Joined: 23-December 09
From: Atlanta, Georgia, USA
Member No.: 8.794



Healthy debate is a good thing smile.gif And so is experimenting with different bits of kit smile.gif In the end you really gotta trust your ears. So if your ears say a recording high bit rate and using third party software to downsample is the way to go then great! If your ears are telling you that it's pretty much the same to start and end at 44.1k, then that's the way to go.

The important thing is to try stuff out and out really see what works best for whos doing the session. Personally I will sometimes go high (96k) and downsample later.

The deliverable is always important since most folks will end up hearing it streamed on soundcloud or youtube. At which point, soooooooo much of our tedious work is "Lost in Translation". Still, it's important to try to have audio as good as possible, just like guitar, it's Noble Journey with no end smile.gif

Todd


--------------------
Go to the top of the page
 
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 


RSS Lo-Fi Version Time is now: 27th March 2017 - 09:16 PM