The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Good dither practices, what are yours? Dynamics Plugins
Old 23rd February 2017
  #331
Gear Guru
 
Karloff70's Avatar
 

ahahahahaha, tube DAW. On more great thing this thread did, it brought stinky out of hiding. Love it.
Old 23rd February 2017
  #332
Lives for gear
 
stinkyfingers's Avatar
 

(...i'll try not to get this too off topic by relating it to dither...)

in the attached pic (or @ that SRC site everyone knows and loves) take a look at Logic's SRC. specifically, v9...
notice the distortion/harmonics ? the terrible 2's...
is it aliasing, is it truncation, is it tubes...i couldn't tell ya.
but what i can tell you is that maybe you should dither before SRC in Logic 9.
dither...it's what's for dinner.

(...looks like with Logic X they yanked the tubes, though,.or at least got everything biased correctly. something like that. i don't know electronics.)
Attached Thumbnails
Good dither practices, what are yours?-logic_src.jpg  
Old 23rd February 2017
  #333
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by stinkyfingers View Post
(...i'll try not to get this too off topic by relating it to dither...)

in the attached pic (or @ that SRC site everyone knows and loves) take a look at Logic's SRC. specifically, v9...
notice the distortion/harmonics ? the terrible 2's...
is it aliasing, is it truncation, is it tubes...i couldn't tell ya.
but what i can tell you is that maybe you should dither before SRC in Logic 9.
dither...it's what's for dinner.

(...looks like with Logic X they yanked the tubes, though,.or at least got everything biased correctly. something like that. i don't know electronics.)
Yup, they did. But watch it, stinky. Don't you come soiling my thread with 'DAWs sound different' type of ish.....next we'll have the null police here and the dither party will be over in a whiffy.
Old 23rd February 2017
  #334
Lives for gear
 
stinkyfingers's Avatar
 

you can't hear the difference between DAWs, thats like being able to hear the difference between dither and truncation @ 24 bit level...

Old 23rd February 2017
  #335
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by stinkyfingers View Post
you can't hear the difference between DAWs, thats like being able to hear the difference between dither and truncation @ 24 bit level...

You're causing a stink, stop it already!
Old 23rd February 2017
  #336
Lives for gear
 
stinkyfingers's Avatar
 

ok, i will be good now...
did you know i used to work for the null police ?
now i work for the D.E.A. (dither enforcement agency)
pay isn't much better, but the benefits are great !

my little dither "secret" or "trick" is this. 99% of what i master has been made with noisy ****ty old analogue stuff (Mook, Bookla, crap like that) and it goes out on vinyl. so what i do is get a noise profile and set my dither just above it, to drown out that awful analog smog. sounds great...
Old 24th February 2017
  #337
Gear Nut
If I'm bouncing a 24 bit 88.2 file to 16 bit 44.1 using the oxford limiter with tpdf dithering,what should I set the dither to? 16 bit or 24. I thought I would bring down the cumulative I.Q. In this thread with a very basic question.
Old 24th February 2017
  #338
Motown legend
 
Bob Olhsson's Avatar
 

Sample rate conversion is signal processing so the 16 bit dither needs to be done afterwards. If you are using a Pro Tools TDM system, you want to set the dither at 24 bits. For native floating point bounces 24 or off are the best settings.
Old 24th February 2017
  #339
Lives for gear
 
Fidelis's Avatar
 

Quote:
Originally Posted by Hyder boy View Post
Can anyone recommend a dither plugin for use with hardware inserts on pro tools 12.
I'm using this one:

https://goodhertz.co/good-dither

It's aax native. Works great! I hope someone do a aax dsp dither plugin for me to use on HDX!
Old 25th February 2017
  #340
Gear Nut
Quote:
Originally Posted by Bob Olhsson View Post
Sample rate conversion is signal processing so the 16 bit dither needs to be done afterwards. If you are using a Pro Tools TDM system, you want to set the dither at 24 bits. For native floating point bounces 24 or off are the best settings.
Thanks for the help here. Please clarify these points if you could.

What is the difference between the way PT HD and native systems handle the process? Why would off be the right setting for native? So confusing.
So are you saying that I should dither the bounced 16 bit file after it's been converted from 24 bit 88.2 to 16 bit 44.1? I fear I have been doing this wrong...
Old 3rd March 2017
  #341
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by seedubs View Post
Thanks for the help here. Please clarify these points if you could.

What is the difference between the way PT HD and native systems handle the process? Why would off be the right setting for native? So confusing.
So are you saying that I should dither the bounced 16 bit file after it's been converted from 24 bit 88.2 to 16 bit 44.1? I fear I have been doing this wrong...
Sounds like you should be converting it to 24 bit 44.1.

And then dithering it to 16 bit 44.1

(some of us swear by doing processing at 32 bit, and then dithering TO 24 bit whenever we make files at 24 bit. I think your question's a 'bit' more basic, though )
Old 3rd March 2017
  #342
Motown legend
 
Bob Olhsson's Avatar
 

TDM systems are fixed-point 56 bit engines having 24 bit connections such as plug-in inputs and outputs. Pro Tools native systems are 64 bit floating point with 32 bit connections so what you need to worry about is the 32 to 24 transition to a converter.
Old 3rd March 2017
  #343
Lives for gear
 

And to be complete, when HD went to HDX (Motorola chips to TI chips): "where HD used 24-bit word lengths for plug-in processing and 48-bit for mixing, HDX uses a 32-bit floating-point resolution for plug-ins and a 64-bit floating-point resolution for mixing."

The 56 bit for TDM is the accumulator per board. Per processing is 48 bit. (both fixed point, obviously) The TDM flex cables between multiple cards were 48 bit (so you suffered a reduction if the per channel processing had to cross boards.)
Old 4th March 2017
  #344
Motown legend
 
Bob Olhsson's Avatar
 

I understand the board crossing problem was solved at some point but the big bottleneck remained the plug-in connections.
Old 8th March 2017
  #345
Gear Maniac
 
Yuri Korzunov's Avatar
 

32 bit integer (and 48 bit too) is not bottleneck. There about -200 dB noise floor.
Old 8th March 2017
  #346
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by Yuri Korzunov View Post
32 bit integer (and 48 bit too) is not bottleneck. There about -200 dB noise floor.
Modern mix topologies are based on compression, saturation and limiting. I can get 200 db gain reduction out of ONE compressor if I pick the right one… and compression ratios multiply, so I could easily get 200 db gain reduction out of three compressors if I'm not careful
Old 8th March 2017
  #347
Gear Maniac
 
arlette's Avatar
What I find strange is that Pro Tools 12 has only dither options for 16, 18 and 20 bit. No 24 bit, even though I record on 24 bit. Can anyone tell me why?
Old 9th March 2017
  #348
Gear Maniac
 
Yuri Korzunov's Avatar
 

Quote:
Originally Posted by chrisj View Post
Modern mix topologies are based on compression, saturation and limiting. I can get 200 db gain reduction out of ONE compressor if I pick the right one… and compression ratios multiply, so I could easily get 200 db gain reduction out of three compressors if I'm not careful
At input of compressor come signal with amplitude more 200 dB,

or

you set input gain inside compressor for achieving this level?
Old 9th March 2017
  #349
Gear Maniac
 
Yuri Korzunov's Avatar
 

Quote:
Originally Posted by arlette View Post
What I find strange is that Pro Tools 12 has only dither options for 16, 18 and 20 bit. No 24 bit, even though I record on 24 bit. Can anyone tell me why?
20 bit quantization moise is about noise level of modern DAC (-120 dB).

24 bit noise level about -140 ... -150 dB.

For 24 bit applying dithering have no sense. Because quantization noise deep drown in noise of DAC.
Old 9th March 2017
  #350
Gear Maniac
 

Quote:
Originally Posted by Yuri Korzunov View Post
20 bit quantization moise is about noise level of modern DAC (-120 dB).

24 bit noise level about -140 ... -150 dB.

For 24 bit applying dithering have no sense. Because quantization noise deep drown in noise of DAC.
Try it on master and listen for space esp. in quality reverbs. You will not hear noise but slightly better separation of sounds. Almost as without dithering it smears or glues more the individual sounds.
Old 9th March 2017
  #351
Lives for gear
 
JP__'s Avatar
 

Once again: Its not about those very low noise levels, its about preventing truncation distortion, which in some cases can becomes obvious through the noise floor) even when it shouldnt (in theory). It depends on quite some different factors. Better trust in practical experience not in theoretical once....
But its always the same; those discussion turns around in circles over and over again always with the same arguments. Its like: everything is already said, but not from everyone....

Better make music.
Old 9th March 2017
  #352
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by Yuri Korzunov View Post
For 24 bit applying dithering have no sense. Because quantization noise deep drown in noise of DAC.
Here's your problem: no such thing as drowning in noise. We hear through noise, we can't hear through truncation (and that's before we even get into fancy wordlength reducers like my NJAD which have insignificant noise compared to TPDF)
Old 9th March 2017
  #353
Here's a nice dither test signal:
http://www.tokyodawn.net/labs/public...tSignal_v1.wav

It consists of:

[impulse left]
[impulse right]

[2 Sine]
[2 Sine -20dB]
[2 Sine -40dB]

[7 Sine]
[7 Sine -20dB]
[7 Sine -40dB]

and then with added brown noise at -60dB to simulate more typical, natural noise content:

[2 Sine*]
[2 Sine -20dB*]
[2 Sine -40dB*]

[7 Sine*]
[7 Sine -20dB*]
[7 Sine -40dB*]

It's easy to analyze via FFT analyzer. Just look for "things" building up between the partials.

I recommend anyone to measure the noisy vs non noisy parts with your favorite dither, and without. My point is, most stereo mixes already have random or sufficiently random content, often 40-50dB (!) above the potential dither noise being added. In many cases, a dither doesn't make a difference (no extra reduction of quantization dist, just a lower SNR).

If homeopathy helps, it's fine! But a waste of time for many. Let's keep things realistic. 200dB of compression on a previously dithered mix is not! It's irrelevant. There's no need to panic.

Further, at these low levels, our brain plays a lot of "music". That's proven at least, called auditive imaginary. Below -100dB, it's far too easy to make things up, it's the expected standard case. Your brain fills the void, the more you think about it, the worse it becomes. That's why objective tests are unavoidable.

Be honest: Can you remember the last record you didn't like because of its quantization distortion?! I can't. if anything, I love the sound of 90's hip hop. Made with good old 16bit or 12bit samplers.

Last edited by FabienTDR; 9th March 2017 at 04:51 PM..
Old 9th March 2017
  #354
Gear Maniac
 

Quote:
Originally Posted by FabienTDR View Post
Here's a nice dither test signal:
http://www.tokyodawn.net/labs/public...tSignal_v1.wav

It consists of:

[impulse left]
[impulse right]

[2 Sine]
[2 Sine -20dB]
[2 Sine -40dB]

[7 Sine]
[7 Sine -20dB]
[7 Sine -40dB]

and then with added brown noise at -60dB to simulate more typical, natural noise content:

[2 Sine*]
[2 Sine -20dB*]
[2 Sine -40dB*]

[7 Sine*]
[7 Sine -20dB*]
[7 Sine -40dB*]

It's easy to analyze via FFT analyzer. Just look for "things" building up between the partials.

I recommend anyone to measure the noisy vs non noisy parts with your favorite dither, and without. My point is, most stereo mixes already have random or sufficiently random content, often 40-50dB (!) above the potential dither noise being added. In many cases, a dither doesn't make a difference (no extra reduction of quantization dist, just a lower SNR).

If homeopathy helps, it's fine! But a waste of time for many. Let's keep things realistic. 200dB of compression on a previously dithered mix is not! It's irrelevant. There's no need to panic.

Further, at these low levels, our brain plays a lot of "music". That's proven at least, called auditive imaginary. Below -100dB, it's far too easy to make things up, it's the expected standard case. Your brain fills the void, the more you think about it, the worse it becomes. That's why objective tests are unavoidable.

Be honest: Can you remember the last record you didn't like because of its quantization distortion?! I can't. if anything, I love the sound of 90's hip hop. Made with good old 16bit or 12bit samplers.
At your respect I think sine wave is not good (enough complex) example of master bus audio :-)
Old 9th March 2017
  #355
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by FabienTDR View Post
If homeopathy helps, it's fine! But a waste of time for many. Let's keep things realistic. 200dB of compression on a previously dithered mix is not! It's irrelevant. There's no need to panic.

Further, at these low levels, our brain plays a lot of "music". That's proven at least, called auditive imaginary. Below -100dB, it's far too easy to make things up, it's the expected standard case. Your brain fills the void, the more you think about it, the worse it becomes. That's why objective tests are unavoidable.
Your brain grasps at straws, all the way from -100dB to -200dB and beyond. Does it pull out a beautiful picture? Or does it pull out a hideous monster? That's why we dither.

When you say 'objective tests' about this situation it's basically abdicating ANY interest in guiding the listener's brain. You don't care, because it's no longer quantifiable. And since you don't care, why does it matter what you think about the subject? We're not listening with FFT analyzers. I use 'em in their place, but dither is about guiding the listener as their brain grasps at auditory straws, so we listen.
Old 9th March 2017
  #356
Quote:
Originally Posted by linkstate View Post
At your respect I think sine wave is not good (enough complex) example of master bus audio :-)
Sine?

A sum of Sines is not a Sine. Take a closer look please, this is a very capable test signal.
Old 9th March 2017
  #357
@chrisj: Of course it's good to care. Dithering has a clear job to do and verifiable benefits. No question about that.

But being too pedantic with it is not worth the time, it is far out of scale. Much like Celsius and Fahrenheit won't mater if we're talking about the temperature of the sun. It is not worth the time to debate about it. Dithering is very important, if not even crucial for lo-fi formats. But beyond 16 bit, it is very hard to demonstrate any advantage. Cost and benefit is my point, it better be in balance, to free time for the truly important decisions (as JP__ pointed out).

Say you're a mastering engineer using a DA AD loop to drive your external gear. This case doesn't need any dithering, anytime, anymore. Simply because the signal already contains more than enough random content. IMHO all this is worth being mentioned: If your signal is naturally noisy, dither won't help. In most cases, your signal will be noisy!

About the perception bias, this is not voodoo, but the basis of very well studied, verifiable mechanisms. I recommend reading and excluding all of these before we talk about -200dB or even -120dB again. Simply because they deliver reasonable explanations of what we perceive and why it doesn't have to relate to the signal at all! Much more reasonable explanations than what I can read above.

https://en.wikipedia.org/wiki/List_of_cognitive_biases

Yes, this list is simply too long. That's why we need objective tests observing how far dithers really affect quantization distortion in a real world signal.
Old 9th March 2017
  #358
Lives for gear
 
JP__'s Avatar
 

As you are naming me here, Fabien; I always use dithering before feeding my analogue chain. After intensive tests and for good reasons. You are wrong here.
Old 9th March 2017
  #359
Wrong with what exactly?

Feel free to try the test signal. Try to pass it through your chain with dither and without. Show me the relevant difference (quantization distortion can only appear in the form of new partials popping up between the test signal, so it's easy to verify). We have both clean and noisy cases in the signal.

Given background noise in the signal above dithering level (very likely the case with music signals, even down to 16 bit), dithering is as wise as insisting on full moon before rendering.

This is a logical conclusion. Not something experienced somehow, somewhere, by someone. Adding noise is the principle of dither, if it's already there, adding more is a waste of time, SNR and resources (not much, though). Simple as that.

Last edited by FabienTDR; 9th March 2017 at 11:32 PM..
Old 9th March 2017
  #360
Lives for gear
 

Quote:
Originally Posted by FabienTDR View Post
@chrisj: Of course it's good to care. Dithering has a clear job to do and verifiable benefits. No question about that.

But being too pedantic with it is not worth the time, it is far out of scale. Much like Celsius and Fahrenheit won't mater if we're talking about the temperature of the sun. It is not worth the time to debate about it. Dithering is very important, if not even crucial for lo-fi formats. But beyond 16 bit, it is very hard to demonstrate any advantage. Cost and benefit is my point, it better be in balance, to free time for the truly important decisions (as JP__ pointed out).

Say you're a mastering engineer using a DA AD loop to drive your external gear. This case doesn't need any dithering, anytime, anymore. Simply because the signal already contains more than enough random content. IMHO all this is worth being mentioned: If your signal is naturally noisy, dither won't help. In most cases, your signal will be noisy!

About the perception bias, this is not voodoo, but the basis of very well studied, verifiable mechanisms. I recommend reading and excluding all of these before we talk about -200dB or even -120dB again. Simply because they deliver reasonable explanations of what we perceive and why it doesn't have to relate to the signal at all! Much more reasonable explanations than what I can read above.

https://en.wikipedia.org/wiki/List_of_cognitive_biases

Yes, this list is simply too long. That's why we need objective tests observing how far dithers really affect quantization distortion in a real world signal.
With all due respect, many here (me included) prefer to dither when it sounds better instead of doing it when it is "needed".

Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump