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
Dithering to 24 bit question (64 bit plugins/DAW) Dynamics Plugins
Old 28th March 2010
  #1
Lives for gear
 
aleatoric's Avatar
Dithering to 24 bit question (64 bit plugins/DAW)

Let me see if I understand this correctly.

When I am working with 24 bit source files and using 64 bit precision plugins (in a 48 bit or 64 bit DAW) I should apply a flat 24 bit dither (no noise shaping) at the end of my chain right? I always bounce processed 24 bit source files to 24 bit and only apply 16 bit noise shaping dither in my assembly DAW (WaveBurner) after any SRC or fades . My understanding is that in this situation I should be applying a flat 24 bit dither when bouncing to 24 bits. However if my final target medium was 24 bits (for DVD audio or archival purposes) I would then use a 24 bit noise shaping dither. Is that making sense? Also, yes I know this is a very minor detail in the grand scheme of things but none the less I would like to confirm that I am indeed doing this "correctly". Thanks.
Old 29th March 2010
  #2
Lives for gear
 
aleatoric's Avatar
No one huh?
Old 29th March 2010
  #3
Lives for gear
 
the unik's Avatar
mmm...When bouncing, that may be correct, though I think it's already done by the plugin itself, processing internaly at 64 bit and resulting in a 24 bits output, not sure though, it may depend on the plugin actually...

AFAIC I don't bounce, just record back the track in realtime into the daw.


Quote:
Originally Posted by aleatoric View Post
Let me see if I understand this correctly.

When I am working with 24 bit source files and using 64 bit precision plugins (in a 48 bit or 64 bit DAW) I should apply a flat 24 bit dither (no noise shaping) at the end of my chain right? I always bounce processed 24 bit source files to 24 bit and only apply 16 bit noise shaping dither in my assembly DAW (WaveBurner) after any SRC or fades . My understanding is that in this situation I should be applying a flat 24 bit dither when bouncing to 24 bits. However if my final target medium was 24 bits (for DVD audio or archival purposes) I would then use a 24 bit noise shaping dither. Is that making sense? Also, yes I know this is a very minor detail in the grand scheme of things but none the less I would like to confirm that I am indeed doing this "correctly". Thanks.
Old 30th March 2010
  #4
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by aleatoric View Post
Let me see if I understand this correctly.

When I am working with 24 bit source files and using 64 bit precision plugins (in a 48 bit or 64 bit DAW) I should apply a flat 24 bit dither (no noise shaping) at the end of my chain right? I always bounce processed 24 bit source files to 24 bit and only apply 16 bit noise shaping dither in my assembly DAW (WaveBurner) after any SRC or fades . My understanding is that in this situation I should be applying a flat 24 bit dither when bouncing to 24 bits. However if my final target medium was 24 bits (for DVD audio or archival purposes) I would then use a 24 bit noise shaping dither. Is that making sense? Also, yes I know this is a very minor detail in the grand scheme of things but none the less I would like to confirm that I am indeed doing this "correctly". Thanks.
It depends on what's going on in your DAW..

If your DAW by default dithers from floating point or 48 bit back into 24-bit integer (which it should, either set by you or by design), then adding an additional dither stage (later or as a plugin on the DAW's output bus, etc), with or without noise shaping is redundant for a 24 bit target medium.

If your assembly program is WaveBurner, I'm assuming your DAW is Logic? If you're going 24 bit file to 64 bit fp process to 24 bit output from Logic (either D/A or bounced file), the conversion from 24 bits to 64 bits and back again dithers down automatically by default in Logic - this internal dither may or may not include noiseshaping (anyone know? either way, ns for >=32 bit floating point to 24 bit integer dither won't reaaally matter, imho).

So the "flat 24 bit dither" you speak of is probably already happening in your DAW, and maybe even w/ noiseshaping included in the process (internally, not sure)... Best to avoid having either of them happen subsequently - alone or together...

Anyone please correct me if I am wrong about this..
Old 30th March 2010
  #5
Lives for gear
 

Dither or no dither at 24 bits: the artefacts would be around -140 dBFS. With the best possible recording systems* the bottom 20-30 dB will be just noise anyway, so why bother polishing the bottom 6 dB of that pile of dust?

*)The very best converters get about 120 dB dynamic range, as do mic preamps. The last 3-4 bits are just random noise always, even if nominally using 24 bits.
Old 30th March 2010
  #6
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by Petrus View Post
Dither or no dither at 24 bits: the artefacts would be around -140 dBFS. With the best possible recording systems* the bottom 20-30 dB will be just noise anyway, so why bother polishing the bottom 6 dB of that pile of dust?

*)The very best converters get about 120 dB dynamic range, as do mic preamps. The last 3-4 bits are just random noise always, even if nominally using 24 bits.
Well, if he's dithering from 24-bit to 24-bit then yes, the dither noise would be inaudible to most humans on most systems.. But it's unnecessary...

Also, most stuff these days will get dithered to 16 bit at some point, so although dithering 64 fp - 24 bit - 24 bit might not make that much of a difference, the extraneous 24 bit dither will become more audibly apparent when that file is dithered to 16 bit.
Old 30th March 2010
  #7
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
My experience has been that failing to dither a 24 bit file can definitely come back around and bite you with crunchy sound upon subsequent processing.

The dithered 24 bit files should sound exactly the same or subtly smoother. Unfortunately sometimes 24 bit software dithering is broken so you always need to use your ears.
Old 30th March 2010
  #8
Lives for gear
 
CZ101's Avatar
 

Quote:
Originally Posted by Bob Olhsson View Post
My experience has been that failing to dither a 24 bit file can definitely come back around and bite you with crunchy sound upon subsequent processing.

The dithered 24 bit files should sound exactly the same or subtly smoother. Unfortunately sometimes 24 bit software dithering is broken so you always need to use your ears.
Interesting.. So you're saying that (let's just use Logic as an example) if you were bouncing to a 24 bit file from Logic, you'd still make sure that there be dithering (that you approved of, sonically) -to- 24 bits going on at the end of the chain.. If so, then I've learned something.

What about the possibility of that 24 bit file being dithered to 16 bit in the future? Would you still put it through 24 bit - 24 bit dither if you thought it might later end up as 16 bit?
Old 31st March 2010
  #9
Lives for gear
 
Coyoteous's Avatar
 

Not to answer for BO, but dither when you reduce word-length... simple as that.

I use flat TPDF, unless I know material will have not further processing... which is never.
Old 31st March 2010
  #10
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
What dithering does is to randomize the bottom bit where the audio cuts off so that the truncation sounds like hiss rather than like distortion. The dither level always needs to be appropriate for the word length you are truncating to and some software developers have screwed this up.

Dither is not used to mask the distortion, it simply eliminates it provided it is done properly.
Old 5th October 2010
  #11
Gear Maniac
 
cjac9's Avatar
Bob,
I'm mixing down in PTLE with a Waves L2 for dithering only. Should I use noise shaping, if so which profile? Moderate, Normal or Ultra?

Thanks,
cjac9
Old 5th October 2010
  #12
Gear Guru
 
UnderTow's Avatar
 

Verified Member
Quote:
Originally Posted by cjac9 View Post
Bob,
I'm mixing down in PTLE with a Waves L2 for dithering only. Should I use noise shaping, if so which profile? Moderate, Normal or Ultra?

Thanks,
cjac9
If this is the final bit reduction to 16 bit, you can use shaping. Use yours ears to decide which one (if any). If you are going to 24 bit, no shaping.

Alistair
Old 5th October 2010
  #13
Lives for gear
 
Cellotron's Avatar
 

Verified Member
In my direct experience there's been no audible difference between the sound of truncating the sum of my DAW's 64bit processing to 24bit versus dithering the same to 24bit. Obviously OMMV! (and this may in fact depend on what processing you apply and what DAW app you use)

Best regards,
Steve Berson
Old 5th October 2010
  #14
Gear Head
 

Quote:
Originally Posted by aleatoric View Post
Let me see if I understand this correctly.

When I am working with 24 bit source files and using 64 bit precision plugins (in a 48 bit or 64 bit DAW) I should apply a flat 24 bit dither (no noise shaping) at the end of my chain right? I always bounce processed 24 bit source files to 24 bit and only apply 16 bit noise shaping dither in my assembly DAW (WaveBurner) after any SRC or fades . My understanding is that in this situation I should be applying a flat 24 bit dither when bouncing to 24 bits. However if my final target medium was 24 bits (for DVD audio or archival purposes) I would then use a 24 bit noise shaping dither. Is that making sense? Also, yes I know this is a very minor detail in the grand scheme of things but none the less I would like to confirm that I am indeed doing this "correctly". Thanks.
Hi,

What is the level of the analog noise floor [e.g. mic pres, hiss, pickup / amp hum, microphones, "component noise", ambient sound,......] in your material?

Is all that quieter than -96dBfs?
Old 5th October 2010
  #15
Lives for gear
 
aleatoric's Avatar
Quote:
Originally Posted by Cellotron View Post
In my direct experience there's been no audible difference between the sound of truncating the sum of my DAW's 64bit processing to 24bit versus dithering the same to 24bit.
+1. This is a somewhat older thread of mine. I have concluded that there is really no need to dither to 24 bits. In my listening tests it proves inaudible. That being said if the final delivery format intended for listening is 24 bits I will apply a noise shaping 24 bit dither. It certainly does not hurt. If anything there could be an audible benefit if DA converters ever started performing at a true 24 bit resolution (144dB dynamic range).
Old 5th October 2010
  #16
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
I'd say it's more often than not inaudible however I've also gotten bit occasionally and for sure the dither is more inaudible if it's working right.

I use the L2 on moderate for 24 bit. or the others with no noise shaping for my mixes. I've occasionally sent pro tools mixes back for a re-bounce using the dithered mixer plug and gotten a less crunchy sounding master.

Samplitude is the only DAW application I've encountered that just handles dither properly and lots of folks say it sounds better. Pro Tools le sounds just as good to me with 24 bit dither using the same converters and monitor system.
Old 6th October 2010
  #17
Gear Head
 

Quote:
Originally Posted by Bob Olhsson View Post
I've occasionally sent pro tools mixes back for a re-bounce using the dithered mixer plug ....
And they bought it?

Not the whole mix, just the bounce?
Old 30th May 2016
  #18
Lives for gear
 
gregor z's Avatar
 

Verified Member
A question from a bit different viewpoint - does it make sense to dither a track that was recorded to analog tape with vintage gear and therefore has a a bit higher then normal amount of noise - which is definitely higher in level then any dither noise would be - so it will be probably 100% masked.

When going from 24 to 24 I guess there is no need for additional dither noise - if I know there will be no further processing down the line.

Maybe even from going from 24 to 16 - there would be no need to add additional noise - if the track in question is "noisy" enough.

Just curious - if dither has any place in tracks that have relatively high self noise.



Thanks in advance,

Gregor
Old 30th May 2016
  #19
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
The dither is less audible than the distortion caused by not dithering. Dither sounds like noise but noise is not dither. If you are using any digital signal processing including simple gain changes, you are no longer going 24 to 24 and need to dither.
Old 3rd June 2016
  #20
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by gregor z View Post
When going from 24 to 24
Is this the case tho?
Technically everything depends on the internal representation of the signal in the daw.
If the daw works internally with some high bit number and you bounce to 24 bits then you need to (again, technically) apply dither.
Wether the source material was only 24 bit and what the bitrate of the processing was is irrelevant in this case i think.

If the internal representation of the signal has more bits than your final output then you technically need to dither.

Edit: Yeah, exactly what you say, Bob. Even a single gain change will 'upgrade' the signal to the internal bitdepth, which will have more than 24 bits. So you need to dither.
Old 4th June 2016
  #21
Gear Maniac
 
Dina Mastering's Avatar
 

I havent researched it, but Id say the 64 bit is not the audio directly, its the processing resolution if that makes any sense. I always thaught its like what happens with a 32bit file, that its nothing more than a 24 bit file with more headroom for peaks, but not more resolution under 0db. I dont think when working with audio at 16 or 24 bits in a 64 bit daw the audio bitrate gets changed or anything, thats why you have to choose a project bitrate anyway, and the audio stays at that bitrate despite the daws "internal resolution".

But appart of the technical stuff, whats most important if it sounds good or not. Do your tests, if it sounds good keep it, if it sounds worse or the same, don't.
Old 4th June 2016
  #22
It's really important to understand that digital audio uses at least two very different forms of representation values. And there are different contexts each having different meanings.

There's the context of an audio app's internal precision. There's the plugins internal precision, and there's also file storage and transfer. All have different implications.


I warmly recommend anyone to grasp this stuff down to the last atom, it's your profession after all. Misunderstanding the canvas is where most trouble begins.

For this specific thread, I wonder why nobody mentioned that we're working with floating point number representation in 99% of the cases. Be it synthesis, processing, mixing, often also storage and transfer (essentially everything except "export for production/distribution").

Now here's an annoying detail:

Floating point signals cannot be dithered (only fixed point signals can).


The quantization distortion produced when floating point signals reach their precision limit is somewhat less problematic than the sound fixed point truncation. Do not try to dither your processing, as it mostly probably bases on floating point representation.

If the source already contains high amounts of noise (i.e. tape, mic, etc), there's really no need to add additional dither before truncation. But it's worth trying out.

Dithering a floating point signal is technically nothing else than an increase of the noise floor. In addition to the quantisation distortion!

Last edited by FabienTDR; 4th June 2016 at 03:43 PM..
Old 4th June 2016
  #23
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by Dina Mastering View Post
and the audio stays at that bitrate despite the daws "internal resolution".
Yeah, that's not really true. As soon as you do any processing the bits of the original file will get mapped to bits in the internal format. Even a small gain change immediately fills up the newy available bits in the internal format. If you then want a 24 bit file at the output again you will need to dither.
Old 4th June 2016
  #24
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by FabienTDR View Post
Floating point signals cannot be dithered (only fixed point signals can).
Huh? Why not? And isn't the mantissa basically an integer (that requires dithering when reduced)?
Old 4th June 2016
  #25
Quote:
Originally Posted by monomer View Post
Huh? Why not? And isn't the mantissa basically an integer (that requires dithering when reduced)?
Floating point numbers are not represented linearly. There's an exponent and a mantissa (you have an exponentiation between them). So adding noise doesn't reduce any quantization distortion, it's just added noise.

http://www.thewelltemperedcomputer.c...tingdither.pdf

Quote:
"In any mathematical system, simple operations create larger numerical results.
Multiplying a 3-digit number times a 3-digit number creates a 6-digit number, for
example. These larger values (more bits, in the digital audio world) create the need for
some form of bit-reduction. Floating-point systems, as well as fixed-point systems, need
to be able to appropriately bit-reduce values. The mathematically correct methodology is
the application of dither, as was first presented and proved decades ago. Dither is easily
and “properly” applicable in fixed-point domains, but it is not so easily applied in
floating-point systems. This paper will explore the issues having to do with dither and
other bit-reduction schemes in floating-point systems."
Old 4th June 2016
  #26
Gear Maniac
 

Quote:
Originally Posted by Dina Mastering View Post
I havent researched it, but Id say the 64 bit is not the audio directly, its the processing resolution if that makes any sense. I always thaught its like what happens with a 32bit file, that its nothing more than a 24 bit file with more headroom for peaks, but not more resolution under 0db. I dont think when working with audio at 16 or 24 bits in a 64 bit daw the audio bitrate gets changed or anything, thats why you have to choose a project bitrate anyway, and the audio stays at that bitrate despite the daws "internal resolution".
But, you should research that first. They are talking about the number of amplitude points irrespective of peak values...
Old 4th June 2016
  #27
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by FabienTDR View Post
Floating point numbers are not represented linearly. There's an exponent and a mantissa (you have an exponentiation between them). So adding noise doesn't reduce any quantization distortion, it's just added noise.

http://www.thewelltemperedcomputer.c...tingdither.pdf
Well, the quote states that it is the mathematically correct to dither floats and fixed point numbers when reducing. So your statement that you cannot dither floats is nonsense according to the text you quote.
Old 4th June 2016
  #28
Gear Maniac
 

Quote:
Originally Posted by monomer View Post
Well, the quote states that it is the mathematically correct to dither floats and fixed point numbers when reducing.
Only when the scaling becomes fixed (floating point to integer conversion).
Old 4th June 2016
  #29
^^ which is nothing else than fixed point
Old 4th June 2016
  #30
Gear Maniac
 

^^ Amen.
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
plus6vu / Mastering forum
41
JAY55 / Mastering forum
20
hackenslash / Mastering forum
4
sound_forward / So much gear, so little time
4

Forum Jump
Forum Jump