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
16bit>32bit float>16bit... Vs 16bit>16bit Dynamics Plugins
Old 28th March 2010
  #1
Gear Maniac
16bit>32bit float>16bit... Vs 16bit>16bit

Hi guys,

Here is the thing:

A.
I have a stereo file in 16bit/44.1Khz format. I have to 'master' it.
My initial thought is to import this in my daw (internally 32bit float), then work my thing and export the result in 16bit.
Would you apply a dither in this case?

B.
Now, in case I have to add another effect my CPU will overload. To avoid this, I will have to export the pre-worked file in 32bit float. Import this file back, continue to work on it, then export it at 16bit.
Would you apply a dither in this case?
Old 29th March 2010
  #2
Lives for gear
 
William Bowden's Avatar
 

Verified Member
Well in case A you will need to convert the 32 bit file (presuming you do processing in 32 bit) to 16 bit. It is normal to add dither when converting down.

In case B if you process the file, export it at 32bit, then further process it at 32 bit, you will only need to dither once at the end when you go to 16bit. If you're asking whether you need to add dither to the 32bit file when you do your first export then I reckon the answer is no.

The King
Old 29th March 2010
  #3
Gear Maniac
In case of A, my original file is 16bit.
My DAW will work it as a 32bit float (internally) but I just import my original 16bit file in my DAW.
Old 29th March 2010
  #4
Lives for gear
 
William Bowden's Avatar
 

Verified Member
Well when I work in Sequoia, even if my orig file is 16bit, as soon as I do any processing eq etc then sequoia displays it as a 32bit file cause the processing is done at 32bit and so for all intents and purposes it 'becomes' a 32 bit file. So if I need to then get it on a cd it has to 'go back' to 16 bit and this is where the dither comes in. I'm not sure what your software is or how it does things, you may be able to do the entire process 16 bit, but sonically you may not find this optimal...

The King
Old 29th March 2010
  #5
Gear Maniac
Well...My DAW is also internaly converting the files to 32-bit float.

You say in any case, we have to use dither to export the 16bit final version?

I always use dither when I convert 24bit into the final 16bit version.
But didn't know I had to use dither when the original source file is 16bit.
Old 29th March 2010
  #6
Lives for gear
 
Adam Dempsey's Avatar
 

Verified Member
Your source resolution is mutually exclusive to your dsp resolution and destination resolution.
So, typically:
source res -> (higher) dsp res (which can include SRC) -> dither & truncate (often part of the same step) to destination res.
Old 29th March 2010
  #7
Gear Maniac
This imply to use dither everytime (when exporting the final 16bit audio), no matter the source file is 24bit, or 16bit... right?
Old 29th March 2010
  #8
Lives for gear
 
William Bowden's Avatar
 

Verified Member
Only if you do some processing (ie if the orig 16bit file doesn't need any processing then leave it 16bit).

The King
Old 29th March 2010
  #9
Gear Maniac
OK.
Thanks for your help!
Old 29th March 2010
  #10
Lives for gear
 
taturana's Avatar
 

Verified Member
if you do any processing the file is not 16 bit anymore... so you do have to convert/dither back to 16bit.
Old 29th March 2010
  #11
Lives for gear
 
CZ101's Avatar
 

You might want to check your DAW's docs and make sure that putting a dithering plugin on the mixbus isn't redundant after the mixer's floating point to integer conversion (which will include dithering)..

Maybe this is all moot, but for example: in Digital Performer there is selectable internal floating point to integer (and vice-versa) dithering which is generally always left on. If the integer bit depth of a project is 16 bit (such as what you're describing here), assuming 16-bit is the desired output word length, in DP, putting a dither plugin on the mixbus, in effect, dithers the output for the second time in a row (obviously not a good thing)..

I've found this topic of dither on the mixbus to be poorly documented in DP and Logic, fwiw..
Old 30th March 2010
  #12
Gear Maniac
I use DP as DAW.
Indeed, there is that Dither function, but for what I read I'd better uncheck this. So I use without the Dither engaged.

I agree, the manual of DP is very poor about this. I have two version: 4.5 and 5.00 both manuals are same about that subject.

But for this project I'm into right now, I will probably use Bias Peak to deal the 32bit-float.

Bias Peak use the POW-R dither, when saving a file you can choose to convert it to 24bit or 16bit. In the dialogue box you also have the possibility to engage the Dither. The weird thing is when your source file is 32bit if you choose to save it to 24bit you can have acces to check the dither function...But if you want to save it as 16bit, the dither case remains greyed. So, you can't apply dither.
The trick is to set an export to 24bit, engage the Dither, then set 16bit export. The Dither function goes grey, but is checked!
I assume this will be applied though.
Old 1st April 2010
  #13
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
I use DP as DAW.
Indeed, there is that Dither function, but for what I read I'd better uncheck this. So I use without the Dither engaged.

I agree, the manual of DP is very poor about this. I have two version: 4.5 and 5.00 both manuals are same about that subject.

But for this project I'm into right now, I will probably use Bias Peak to deal the 32bit-float.

Bias Peak use the POW-R dither, when saving a file you can choose to convert it to 24bit or 16bit. In the dialogue box you also have the possibility to engage the Dither. The weird thing is when your source file is 32bit if you choose to save it to 24bit you can have acces to check the dither function...But if you want to save it as 16bit, the dither case remains greyed. So, you can't apply dither.
The trick is to set an export to 24bit, engage the Dither, then set 16bit export. The Dither function goes grey, but is checked!
I assume this will be applied though.
I agree - poor explanation in an otherwise pretty good manual.

Anyways - by all accounts leave the dither option checked in DP. From what I understand, the option enables internal dithering for gain adjustments as well as for floating point to integer conversion (I think ). Apparently, the "dither" option (with the checkmark - in one of the main pull down menus - not the bounce dialogue box) does not have anything to do with bounces being dithered to reduced bit depths.
Old 24th April 2010
  #14
Gear Maniac
Ok guys,

I'm back with my Dither dilemna...
The original file I'm working on is 44.1Khz/16bit. I've been told it has been recorded in 16bit but I doubt it as the spectrum analyzer clearly shows the dither print on the high freqs.
I have mastered my audio in DP an exported the mastered sound as a 44.1Khz/32bit float file.
Then I import this file into Bias Peak, to export it as a 16bit file.
That where I apply my dither (powR3).

The result (16bit dithered file) sounds a little bit 'shh-shh' to me.
And I'm considering to not apply dither on this one.

Please, note that I would do on anything recorded in 24 bit but it looks this file has already been dithered before being sent to me.

I have converted my 32bit file into 16bit without dither, and I don't hear the 'dither on dither' artifact sound.

What would you do?
Old 24th April 2010
  #15
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
Ok guys,

I'm back with my Dither dilemna...
The original file I'm working on is 44.1Khz/16bit. I've been told it has been recorded in 16bit but I doubt it as the spectrum analyzer clearly shows the dither print on the high freqs.
What spectrum analyzer? How could you possibly determine that what you are seeing in the analyzer is dither noise as opposed to the myriad other possibilities?

Quote:
Originally Posted by _barnee View Post
I have mastered my audio in DP an exported the mastered sound as a 44.1Khz/32bit float file.
Ok - with the "dither" option checked or not in DP?

Quote:
Originally Posted by _barnee View Post
Then I import this file into Bias Peak, to export it as a 16bit file.
That where I apply my dither (powR3).

The result (16bit dithered file) sounds a little bit 'shh-shh' to me.
And I'm considering to not apply dither on this one.
Every time you play the 32-bit file, either in DP or Peak, it gets crunched down to 24 bits integer (in one way or another - dither or none) so that your converters can do their job. My guess is that Peak dithers 32-bit float on playback to the converter bit-depth by default, but I could be wrong.

It sounds like you need to once and for all figure out how many times dither is being applied after you import the original file into DP.. Don't worry if it's already been dithered to 16-bit.. Even if it has, is there anything you can do to change that downstream? If you, in your processing are adding redundant dither stages, then this could be a problem, but then again, all of this very subtle stuff can really play havoc with our bias expectations, etc...

Quote:
Originally Posted by _barnee View Post
Please, note that I would do on anything recorded in 24 bit but it looks this file has already been dithered before being sent to me.
It would be a good idea to dither from 32-bit float to 24 bits in that case, yes. But it's even more important from 32-bit float to 16 bits for obvious reasons - regardless of whether it had already been dithered to 16 bits..

Quote:
Originally Posted by _barnee View Post
I have converted my 32bit file into 16bit without dither, and I don't hear the 'dither on dither' artifact sound.

What would you do?
Are you absolutely sure there was no dither (hidden, default, whatever) when you converted from 32-bit float to 16-bit integer in this case?
Old 24th April 2010
  #16
Lives for gear
 
Waltz Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by _barnee View Post
I have converted my 32bit file into 16bit without dither, and I don't hear the 'dither on dither' artifact sound.
I would go with what ever sounds best to you. Truncation or dither.

The general consensus is to dither when ever reducing the word length...even if you are receiving or working on a file that's has already been dithered and in turn work on that file at a higher bit depth.

Depends on style of music, type of dither and if you can actually hear and or prefer different types or none at all.
Old 24th April 2010
  #17
Gear Maniac
Hi guys, thanks for your replies.

There is no chance I can get another mix.

Well...How i know about the dither spectrum signature?
I did several experiences in the past, and especially on low levels you can really see a dither signature. It is typical.

How i know about 'my' last conversion does not include dither?
I don't enable the option when exporting/saving my file.

Yep, the dither option is checked in DP.
But is that a trouble?
Old 25th April 2010
  #18
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
Yep, the dither option is checked in DP.
But is that a trouble?
No, that's good.
Old 25th April 2010
  #19
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
My experience has been that every time dither sounded worse, either I'm still monitoring 24 bits and the audio isn't being truncated until its written to a file or disk or else the software's dither routine turned out to be defective.

When its working right, you shouldn't hear it at all or it should sound smoother and less edgy.
Old 25th April 2010
  #20
Gear Maniac
I have no trouble with dither all audio I had in the past, from 24bit to 16bit.
But this is the first time I work on a 16bit mix, which obviously has already been dithered.

I am a bit confused, because there is "apply dither when you downsize to 16bit" and in this case: this sounds weird once applied.
As suggested Waltz, should I just use truncation instead?

The dithering process in Bias Peak is applied only during saving the file.
Once you choose to save the file, a box opens and ask you to name the file and locate it on your drive. But it also lets you choose the file format (AIF, WAV, etc...), the bit depth and if you want to apply POW-R.

My next test will be to carefully listen to the audio when I engage a maximizer/limiter plug in with dither option. Just to be sure about the result (double dither artifact).
Old 25th April 2010
  #21
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
I have no trouble with dither all audio I had in the past, from 24bit to 16bit.
But this is the first time I work on a 16bit mix, which obviously has already been dithered.

I am a bit confused, because there is "apply dither when you downsize to 16bit" and in this case: this sounds weird once applied.
As suggested Waltz, should I just use truncation instead?

The dithering process in Bias Peak is applied only during saving the file.
Once you choose to save the file, a box opens and ask you to name the file and locate it on your drive. But it also lets you choose the file format (AIF, WAV, etc...), the bit depth and if you want to apply POW-R.

My next test will be to carefully listen to the audio when I engage a maximizer/limiter plug in with dither option. Just to be sure about the result (double dither artifact).
One thing that might help elucidate what's happening in your situation here is a "bitscope"-type application:

Bitter
Bitscope DSO

My guess is - Peak is (automatically/by default) dithering from 32-bit float to 24 bit integer when you play back the file (you have your Peak project bit depth set to 24 bits I'm assuming).. You can check for this with the applications listed above - whether Peak is automatically dithering to 24 bits or 16 bits before D/A conversion...

When Peak gives you the option to export to 16 bit and to add dither in the bounce process, I wonder if selecting dither at that stage in effect is adding it a second time in a row - first (by default/automatically) from 32-bit fp to 24 bit integer (or 16 bit depending on the project bit depth) - then once more by your selection - to the bounced 16-bit file...?

If this is the case (and it very well might not be, so this could all be moot), it might make more sense, as counter-intuitive as this seems, to set your Peak project bit-depth to 16 bits, but I'm not sure as I don't own Peak..
Old 25th April 2010
  #22
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Virtually all signal processing destroys the dither including even the slightest gain changes.

It's really really stupid that all software doesn't simply handle dithering when it's needed. It is nothing short of utterly ignorant, incompetent or lazy programming. People shouldn't ever need to be asking about using dither.
Old 25th April 2010
  #23
Lives for gear
 
Greg Reierson's Avatar
 

Verified Member
Quote:
Originally Posted by Waltz Mastering View Post
The general consensus is to dither when ever reducing the word length...
Not really a consensus. More of a mathematical fact, unless distortion is your goal.

A friend asked me about dither just yesterday. I dug up an old Dan Lavry post. Worth a read - especially the section on TPDF.

GR

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

From Dan Lavry:

When the signal gets to be low, and we “run out of” quantization levels (not enough bits relative to small signal), the error signal becomes signal dependent - no more a random noise behavior. Part of the error signal becomes harmonic distortions, and the rest does becomes noise, but not at constant amplitude. The noise “going up and down in level” is called “noise modulation”.

Of course, adding random noise (dither) is a compromise, but just slight increase in the noise floor is a small price to pay. The harmonic distortion and noise modulation are a lot more offensive. Yet, we are interested in adding as little noise energy as possible.

Dr. Lip****z and DR. Vanderkooey from Toronto university presented the solution, which is TPDF (triangle dither). They found out that 1 LSB rectangular dither is enough to eliminate the harmonic distortions, but the noise modulation does not disappear.

Say you have a quantization levels at -2,-1,0,1,2,3… and the dither amplitude is .999 (near 1). With zero signal, if the DC is at 0, the dither will be between 0 and 0.999, never hitting 1, so the output code is always at zero. No noise at all. But say we move the DC to +0.5. Now the dither is between 0.5 and 1.499. So about half the time we are between 0 and 1, the rest of the time we are between 1 and 2. We now have noise.

If a tiny bit of DC can change the noise, so can a small signal… That is how noise modulation occurs. So Lip****z and Vanderkooey figured that the minimum noise energy required to eliminate both distortions and noise modulation is triangle noise (statistical distribution wise). The idea is to have the same constant noise level on average (changing DC or adding a small signal should not effect the noise amplitude.

What is triangle probability, how do we make it?

Lets, begin with rectangular probability. We take a dice and throw it many times, and soon enough we learn that each number (1 through 6) is about as likely as the other. With a lot of dice throwing we find equal probability thus a rectangular shape

1 x
2 x
3 x
4 x
5 x
6 x


Now let’s throw 2 dice. We can get a 1+1=2, or a 3+4=7 or “anything”. But there is only one way to get a 2 - (1+1)
two ways to get a 3 – (1+2) or (2+1)
three ways to get a 4 – (1+3) or (3+1) or (2+2)
four ways to get a 5
five ways to get a 6
six ways to get a 7
five ways to get a 8…
one way to get a 12 (6+6)

Combined to a picture it looks like a triangle:

2 x
3 x x
4 x x x
5 x x x x
6 x x x x x
7 x x x x x x
8 x x x x x
9 x x x x
10 x x x
11 x x
12 x

There are 2 requirements to get (on average) a triangle shape:
1. Each dice is random or “fair”
2. The dice do not influence each other, they are independent.

Therefore generating a triangular dither is done by adding 2 separate and INDEPENDENT rectangular dithers, 1LSB in amplitude each. The minimum value is 0+0=0. The maximum value is 1+1=2 LSB’s, and the energy is concentrated (peak of the triangular) at 1.

If you make that dither and subtract 1, you end up with:
The minimum value is 0+0-1= -1. The maximum value is 1+1-1= 1 LSB’s, and the most of the time (peak of the triangular) you are near 0, thus most of the time you are not adding too much noise… The main point is that if you now add DC (or signal), the dither energy stays constant(I am not explaining in detail but it is so).

Dan Lavry

----------------------------------------------
Old 25th April 2010
  #24
Gear Maniac
Thanks for the Bitter plug in, I'll test it soon.

Bias Peak just 'follow' the file it opens. There is no bit depth settings.
There is a bit usage function, and it shows 31 bit used.
Will check with the Bitter plug.
Old 25th April 2010
  #25
Lives for gear
 
Waltz Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by Greg Reierson View Post
Not really a consensus. More of a mathematical fact, unless distortion is your goal.
Thanks for the link, I understand this. I was speaking in very "general" terms (especially regarding re-dithering) in the hopes of avoiding another endless thread about dither.

Quote:
Originally Posted by _barnee View Post
As suggested Waltz, should I just use truncation instead?
I was not suggesting this. Merely pointing out that this is the option of not using dither.

I would always use dither when reducing bit depth.

Quote:
Originally Posted by Bob Olhsson View Post
It's really really stupid that all software doesn't simply handle dithering when it's needed. It is nothing short of utterly ignorant, incompetent or lazy programming. People shouldn't ever need to be asking about using dither.
I concur.
Old 25th April 2010
  #26
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
Thanks for the Bitter plug in, I'll test it soon.

Bias Peak just 'follow' the file it opens. There is no bit depth settings.
There is a bit usage function, and it shows 31 bit used.
Will check with the Bitter plug.
Is there any reason you have to go to Peak with the 32-bit float file? It might make more sense just to stay in DP because actually - DP processes its master bus in 64-bit float by default..
Old 25th April 2010
  #27
Gear Maniac
I import the file in Bias Peak to use the POW-R dither.
But I'm might consider using Ozone dithering. Never used it for a big while since I have Bias Peak, but maybe I will get better result with it.
I've read that Ozone dither was one of the best circulating (plug in speaking).

Any thoughts?
Old 25th April 2010
  #28
Gear Maniac
Got news.

The following pictures shows:
  • the original file (44.1Khz/16bit), in Digital Performer.
  • the mastered sound file 32bit float, in Bias Peak.
  • the mastered sound file downsized to 16bit with POWR3 dither, in Bias Peak.
  • the mastered sound file downsized to 16bit with no dither, in Bias Peak.

Have tested Bias Peak and Digital Performer with a wav coming from a CD (pressed), same thing.

2 options:
  1. Bias peak is...flawed?
  2. the Bitter plug-in does not like Bias Peak.

Something just came to mind, I'll try to import the dithered/no dithered file in DP just to see what Bitter says.
Attached Thumbnails
16bit>32bit float>16bit... Vs 16bit>16bit-32bit-float.png   16bit>32bit float>16bit... Vs 16bit>16bit-16bit-dither.png   16bit>32bit float>16bit... Vs 16bit>16bit-16bit-nodither.png   16bit>32bit float>16bit... Vs 16bit>16bit-original-file-dp.png  
Old 25th April 2010
  #29
Gear Maniac
Okayyy...

I've just relaunched Bias Peak, and this time used the Bit Usage function.
On both 16bit (dither and no dither), it shows the files are indeed 16bit.
So, it looks like Bitter does not like Bias Peak.

I've read Bitter was supposed to be useful and tell if a Dither was applied or not... But I really don't see how I could find this in the plug-in.
It only shows if there are clippings and intersample clippings (which is fine already).
Old 25th April 2010
  #30
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by _barnee View Post
Okayyy...
I've just relaunched Bias Peak, and this time used the Bit Usage function.
On both 16bit (dither and no dither), it shows the files are indeed 16bit.
So, it looks like Bitter does not like Bias Peak.
Does Bitter's meter now display the same thing as the images you posted above or does the meter look different?

Quote:
Originally Posted by _barnee View Post
I've read Bitter was supposed to be useful and tell if a Dither was applied or not... But I really don't see how I could find this in the plug-in.
It tells you if your bit-depth is actually being reduced or not; and if there's truncation or dither...

The dither/no dither readings in the pics are confusing, to say the least..

Let me get a few things straight -

You put "Bitter" as the last thing in Peak's plugin chain, yes?

You got these pics while the audio played in real-time, I suppose.. Is the Pow-R dither applied in the form of a plugin or as a setting/option inside Peak?

Lastly, do you own Audio Hijack Pro or a similar program that can receive an audio stream from Peak?

I ask because it seems like you need to check Peak's audio output with Bitter active in another program - away from Peak - because in the pics it's registering as 32-bit float for both your "dithered" and "non-dithered" pics, unless I am mistaken.. It seems like that is either a problem with Peak or with Bitter, but I'm not sure which..

But if Bitter is working, then after your audio is "dithered" (wherever that is happening in the chain in Peak) - Bitter shouldn't display any activity above the target bit-depth... Then again, Bitter could be displaying Peak's 16-bit integer result (dithered or not) but still formatted for 32-bit float..?
Top Mentioned Products
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
Similar Threads
Thread
Thread Starter / Forum
Replies
zzz / Mastering forum
3
m_gant / Mastering forum
14
Beezoboy / Music Computers
9
A1A2 / So much gear, so little time
7

Forum Jump
Forum Jump