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 13th March 2017
  #481
Quote:
Originally Posted by qwe View Post
If your goal is the very best possible sound quality, then you've got to look at all details, big and small.
Sure, I like the idea of a minimal signal path.
Old 13th March 2017
  #482
Airwindows
 
chrisj's Avatar
Well, we have some things in common at least. Peace? (though I hadda give up drugs more than twenty years ago)

The thing is, when it gets down to scientific method, we've got some authorities and they ain't me, or you. I defer to them, and I've been in some lively arguments where they were present. I think I mentioned JJ… we ended up agreeing, in one Gearslutz thread, that the threshold of ultimate sonic infinity was in the area of 20 or 22 bits, 64K or so sampling rate. Done properly (and I mean PROPERLY, dithering and everything just so), no human being was likely to be able to require more, even at their most perceptive moments.

Of course it's more convenient to process 24 or 32 bit words, and of course we've already got 96K as an established format. Also, this is in reference to an output format, not intermediate formats (a video equivalent might be Apple ProRes or motion RAW: while it's possible to fool the eye with video compression, processing it can lead to serious degradation).

I guess I just feel I know how to make subjective experience fireworks explode better. Which is not always a blessing: I've sometimes been able to deliver extremely vivid 'sound pictures' of ugly or confused music, which offended the listeners so much that they insisted I must quit and concede the field to real musicians! That's some kind of strange victory, being able to annoy unfriendly music listeners that badly. All my auditory tricks collaborated in getting this striking, visceral result

We're not done with digital audio theory. It's not a settled question. Hell, for years people didn't really understand what jitter did: now, the role of dither, the relevance of truncation in every math operation at 32-bit floating point (also 64 bit of course, but much like JJ with the 20/64K there comes a point where you quit worrying about it)… these things are changing how we approach digital audio.

Maybe we'll even be able to sell it, if we de-crapify it enough
Old 13th March 2017
  #483
Gear Maniac
 
Yuri Korzunov's Avatar
 

Quote:
Originally Posted by qwe View Post
I'm confused--if you're sending the DAC 16-bit data as a 24-bit stream, then it's not switching mode?

Or do you mean that, by zeroing out the 8 LSBs it will have an impact on the DAC's operation like delta-sigma noise shaping loops?
I like that you suggest. It is very resonable.

I’m not sure that any DAC apply left shifting 16-bit with zeroing out 8 lesser bits.

Also I’m not sure that any DAC provide 0 … 0.1 dB difference levels in 16 and 24 bit mode, that need, in my opinion, for avoiding difference in subjective perception of quality.

But! We can convert 16 bit file to 24 bit by shifting 16-bit with zeroing out 8 lesser bits. It is lossless.
And compare two 24 bit files at any DAC: with 8 LSB zeroes and without.



However, I see other technical issue for this way.

Let suggest that work only upper 16 bit. Such mode have level distortion that defined by precision of these 16 bit (I think, anyway DAC contains resistors with certain error of value).

When we use 24 bit mode, begin work group of lesser 8 bits. This group add own distortions. And total level of distortion is changed.

So even for way, that you suggest, we compare not formats, but DAC implementation again.
Old 13th March 2017
  #484
qwe
Lives for gear
 

Quote:
Originally Posted by FabienTDR View Post
Quote:
Originally Posted by qwe View Post
If your goal is the very best possible sound quality, then you've got to look at all details, big and small.
Sure, I like the idea of a minimal signal path.
Nah, "minimal" signal path isn't "best possible sound quality," that's a "minimalist"/so-called "purist" approach. It's meaningless to an extent to make a division between what's generated "at source" and what's done in post-processing; for example, pianos have their own "exciter" built in--the undamped strings of the upper octaves.

If psychoacoustics tricks are "bad," then better remove them from the instruments as well and use only tuning forks or sine wave generators.

For example, to do a "micro pitch shift," given the choice between Eventide Quadravox (H8000-derived algorithm) and the "emulations" of the old Eventide gear (e.g. H910), I'd use Quadravox.

The idea that either you're doing a "purist" recording (in which case keep processing to a minimum) or you're doing an "impure" recording (in which case anything goes?) is surely a frequent cause of medicore sound quality...

At the same time, some of the "audiophile" recordings out there are atrocious...
Old 13th March 2017
  #485
qwe
Lives for gear
 

Quote:
Originally Posted by chrisj View Post
I am MORE than happy to spend days and weeks and months and years working out what would best condition digital audio for my Neil Young 'Pono', which is an AMAZING little portable audio player with known and ingenious D/A conversion (it does some unorthodox things given high sample rate). I'll do this knowing perfectly well that almost nobody has a Pono.
Doesn't the Pono use the ESS9018 DAC? What is it doing that's "unorthodox"?


Quote:
Originally Posted by chrisj View Post
Most people have iPhones or cheap Android phones and earbuds. I don't have to care about that, and I don't. The audio I (and my compatriots) produce, is there for if those people want to get all audiophile for a hobby.
Why on earth would anyone using this forum ever, or at least do any serious listening, using cheap earbuds?
Old 22nd April 2017
  #486
What an exciting thread and topic; I hope it doesn't stop here.

As a mere mixing and mastering engineer, watching the "fierce discussion" between two of my admired devs, Fabien Schivre (TDL) and Chris Johnson (Airwindows), was astounding; not to mention the early posts by Bob Olhsson. We learnt such a lot.

The thing that's not too good though is that we have, or I personally have, not reached a definitive conclusion. This is dithering we're talking about, so I'm not sure the approach of "let's try this myself at the studio and if I hear a change or I like it better with dither, then it stays" will be the right way to go.
Old 22nd April 2017
  #487
qwe
Lives for gear
 

Quote:
Originally Posted by morfi View Post
The thing that's not too good though is that we have, or I personally have, not reached a definitive conclusion. This is dithering we're talking about, so I'm not sure the approach of "let's try this myself at the studio and if I hear a change or I like it better with dither, then it stays" will be the right way to go.
Unless you have any reason *not* to use it, then why wouldn't you...?
Old 23rd April 2017
  #488
Quote:
Originally Posted by qwe View Post
Unless you have any reason *not* to use it, then why wouldn't you...?
I've been using it, I've been hearing it (or at least I think I have =D ) and it's been all for the better...

...until Fabien, whom I respect such a lot after reading his articles and specially enjoying the out-of-ordinary sound quality of his plug-ins, said that I should be wrong or shouldn't been hearing anything.

Of course he can be wrong just like anybody but man, a statement from this guys just makes you think.
Old 23rd April 2017
  #489
I don't see much problem with adding dithering noise, I'm just curious about any practical proofs in the general sense. We'll maybe release a small free truncation monitor allowing to A/B both truncation results, 2lsb error and further an automatic check for randomness purity of that error. Maybe that helps bringing some objectivity into the decision.
Old 23rd April 2017
  #490
@FabienTDR that would make for a very interesting tool. Could be something just like the Ultrasonic filter you released some time ago (albeit that one didn't ever leave the alpha stage )

I guess that the coming "Limiter No.7" from TDL won't include a dither stage then. That's not a problem, but it does make me wonder about the subject when I consider the inclusion of a dither built-in in other geniouses' brickwall limiters like Paul Frindle's Sonnox Limiter and DSM, Atusto Kubo's Invisible Limiter G2, Dave Gamble's Limitless and maybe one or two more.
Old 23rd April 2017
  #491
qwe
Lives for gear
 

Quote:
Originally Posted by qwe View Post
Unless you have any reason *not* to use it, then why wouldn't you...?
I should really add: As long as it's flat TPDF dither.
Old 24th April 2017
  #492
Deleted User
Guest
Holy cow, thanks for this thread! Like probably most people, I had no idea you were supposed to add dither to every DAC output. Makes sense. Got some TPDF on there and it definitely sounds better!
Old 4th November 2017
  #493
Gear Head
 

Hi All:

I’m mixing a big band jazz album, and plan to deliver 32 bit files out of Digital Performer to the Mastering Engineer. My question is: When bouncing 32 bit mixes out of DP, which is 32 bit float, but using 64 bit plugins on the stereo bus, should I use 24 bit dither to dither down from 64 bit to 32 bit? Or should I just not dither at all? I'm guessing that dithering from 64 bit float to 32 bit float is not necessary. Any opinions?

Thanks,

Robert Cartwright
perfectwavedigital.com
San Diego
Old 4th November 2017
  #494
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by rcartwright View Post
Hi All:

I’m mixing a big band jazz album, and plan to deliver 32 bit files out of Digital Performer to the Mastering Engineer. My question is: When bouncing 32 bit mixes out of DP, which is 32 bit float, but using 64 bit plugins on the stereo bus, should I use 24 bit dither to dither down from 64 bit to 32 bit? Or should I just not dither at all? I'm guessing that dithering from 64 bit float to 32 bit float is not necessary. Any opinions?

It's possible but that's not how you'd do it. Also, '64 bit plugins' might mean how much memory space they can address: the stereo buss is going to be either 32 bit already, or 64 bit (double) which some DAWs can do. If you mean the big transition from 32 bit to 64 bit plugins, that doesn't relate to how the buss is, that relates to how much RAM the plugin can address, and it may well be a 32 bit buss anyway.

I'm gonna say, don't use 24 bit dither unless you're going to output at 24 bit fixed point.
Old 4th November 2017
  #495
Gear Head
 

Quote:
Originally Posted by chrisj View Post
It's possible but that's not how you'd do it. Also, '64 bit plugins' might mean how much memory space they can address: the stereo buss is going to be either 32 bit already, or 64 bit (double) which some DAWs can do. If you mean the big transition from 32 bit to 64 bit plugins, that doesn't relate to how the buss is, that relates to how much RAM the plugin can address, and it may well be a 32 bit buss anyway.

I'm gonna say, don't use 24 bit dither unless you're going to output at 24 bit fixed point.
Hi Chris:

By 64 bit, I don't mean as in a 64 bit OS or app, but a double-precision architecture within the plugin. Therefore, 32 bits in the DAW with 64 bit precision within the plugin. I agree that it seems like adding dither to bounce to a 32 bit file is probably unnecessary.

Thanks!

Robert
Old 5th November 2017
  #496
Lives for gear
 
Arksun's Avatar
Quote:
Originally Posted by rcartwright View Post
Hi All:

I’m mixing a big band jazz album, and plan to deliver 32 bit files out of Digital Performer to the Mastering Engineer. My question is: When bouncing 32 bit mixes out of DP, which is 32 bit float, but using 64 bit plugins on the stereo bus, should I use 24 bit dither to dither down from 64 bit to 32 bit? Or should I just not dither at all? I'm guessing that dithering from 64 bit float to 32 bit float is not necessary. Any opinions?
Dither is meant to work by adding the noise before truncation, not after. So using a dither plugin to try and dither the 64-bit signal down to 32-bit in a DAW that runs at 32-bit float mixer engine would not be possible unless it was actually inside the plugin itself.
Old 5th November 2017
  #497
Lives for gear
 
toledo3's Avatar
 

Quote:
Originally Posted by Deleted User View Post
Holy cow, thanks for this thread! Like probably most people, I had no idea you were supposed to add dither to every DAC output. Makes sense. Got some TPDF on there and it definitely sounds better!
Wow, is this where the conversation went?

Nope.

Regardless of what anyone in this thread says, in generalities... the best bet is:

READ THE MANUAL.

If your DAW software doesn't say that you need to add dither after DSP chains, for example... YOU DO NOT NEED TO.

Many DAWs (most?) add TPDF dither at the end of the internal mixer. If you're working in 24 bit, even if it steps up to 32 for DSP, or 64 for the internal mixer, it is giving you a non noise shape dithered 24 bit output. Anything else is atypical.

People who want to sell noise shaped dither products or similar may purposely try to obsfucate reality.

Post processing can sometimes sound nasty when tracks or a mix has had one or more rounds of unnecessary noise shape dither applied. So best to avoid doing it when not needed, and *best* to read the manual for your software and gear to TRULY understand your own system inside and out, because every so often there are quirky implementations.
Old 5th November 2017
  #498
Deleted User
Guest
Quote:
Originally Posted by toledo3 View Post
Wow, is this where the conversation went?

Nope.

Regardless of what anyone in this thread says, in generalities... the best bet is:

READ THE MANUAL.

If your DAW software doesn't say that you need to add dither after DSP chains, for example... YOU DO NOT NEED TO.

Many DAWs (most?) add TPDF dither at the end of the internal mixer. If you're working in 24 bit, even if it steps up to 32 for DSP, or 64 for the internal mixer, it is giving you a non noise shape dithered 24 bit output. Anything else is atypical.

People who want to sell noise shaped dither products or similar may purposely try to obsfucate reality.

Post processing can sometimes sound nasty when tracks or a mix has had one or more rounds of unnecessary noise shape dither applied. So best to avoid doing it when not needed, and *best* to read the manual for your software and gear to TRULY understand your own system inside and out, because every so often there are quirky implementations.
Hah, indeed - it's where I left the conversation but about a week later I reverted and decided no dither sounds better. I'm sure the Cubase manual does not include important details like this though.
Old 6th November 2017
  #499
Gear Head
 

Quote:
Originally Posted by Arksun View Post
Dither is meant to work by adding the noise before truncation, not after. So using a dither plugin to try and dither the 64-bit signal down to 32-bit in a DAW that runs at 32-bit float mixer engine would not be possible unless it was actually inside the plugin itself.
Yes, that's very true. The issue is resolved for me - thanks, all!

Robert
Old 15th December 2017
  #500
Gear Maniac
 

ive been busy a couple of days with analysing my system and reaper. also tryd out some different dithers and i am glad i did!
my favs right now are paulDither and Kazrogs masterDither (the latter does seem to be individualy randomized between L and R)
reapers dither plugin is also ind randomized LR but doesnt serve nearly as good as the once above.

theres a difference between them, paulDither truncats to 24bit while masterDither outputs 32bit (floating i guess)

is the 32bit output from the masterDither an issue in any way?

ill be using dither when taking an analogue trip, on the monitor buss and for rendering to lower depth and fixed point.
Old 15th December 2017
  #501
Motown legend
 
Bob Olhsson's Avatar
 

Quote:
Originally Posted by toledo3 View Post
...Many DAWs (most?) add TPDF dither at the end of the internal mixer. If you're working in 24 bit, even if it steps up to 32 for DSP, or 64 for the internal mixer, it is giving you a non noise shape dithered 24 bit output. Anything else is atypical...
Can you list the ones that do this? Samplitude is the only one I'm aware of. Logic has a dither option for bouncing but they are all noise shaped. Pro Tools has none and requires a 3rd party plug-in. (I've been using GHz Good dither lately.) A fair number of developers don't seem to understand dither which is scary to someone like me who is merely a user.
Old 15th December 2017
  #502
Motown legend
 
Bob Olhsson's Avatar
 

Your system will truncate 32 to 24 so it's not really a problem. I'll have to re-check master dither.
Old 15th December 2017
  #503
Gear Maniac
 

Thanks bob!
Masterdither is by far the widest of them all. Maybe too wide for some styles...

How is the goodhertz u think?
Old 15th December 2017
  #504
Motown legend
 
Bob Olhsson's Avatar
 

By definition dither should be completely random between channels.
Old 15th December 2017
  #505
Lives for gear
 
Taurean's Avatar
Quote:
Originally Posted by Bob Olhsson View Post
Can you list the ones that do this? Samplitude is the only one I'm aware of. Logic has a dither option for bouncing but they are all noise shaped. Pro Tools has none and requires a 3rd party plug-in. (I've been using GHz Good dither lately.) A fair number of developers don't seem to understand dither which is scary to someone like me who is merely a user.
Sonar (RIP lol) had the option to turn on intermediate dither for this purpose. Since you can choose bit depths for recording, importing, and bouncing, freezing, and "applying effects" function. As well as dither on bounce down. The dither options were POW-r 1-3, rectangular and triangular.
Old 15th December 2017
  #506
Gear Maniac
 

Quote:
Originally Posted by Bob Olhsson View Post
By definition dither should be completely random between channels.
yea i heard you write that somewhere. when checking with span, the phase cor meter is at 0 with kazrog masterDither so L and R should be randomized between each other.

i like in reaper where the monitor buss is free for adding dither or whatever you want. also applying pan law where ever you want is possible. before or after the monitor buss, if that would make any difference i dont know but i feel better with the dither absolutly last in the system.
Old 15th December 2017
  #507
Motown legend
 
Bob Olhsson's Avatar
 

It needs to be last or else it can't prevent the distortion. This is part of the math What is dither? | EarLevel Engineering

While I'm here (excuse my geekiness...) Sampling theory, the best explanation you’ve ever heard—Part 1 | EarLevel Engineering

Interestingly he thinks 24-bit dither is unnecessary. That suggests limitations in the quality of his monitoring to me. I only hear it sometimes but I've never heard it do anything bad and I've found truncation distortion buildup to be a real problem. If it's a pain in the ass, the world won't end if you don't dither but when it's trivial, it's pretty stupid not to.
Old 15th December 2017
  #508
Gear Maniac
 

and then one starts to think about who is making the truncation(DA, firmware, DAW...) and if one is better than the other.

thanks for the links Bob. geek on!
Old 15th December 2017
  #509
Lives for gear
 

Quote:
Originally Posted by 5.333V View Post
and then one starts to think about who is making the truncation(DA, firmware, DAW...) and if one is better than the other.

thanks for the links Bob. geek on!
True truncation (fixed point to fixed point) is done only one way (dropping the bits — the name truncation is exactly what it sounds like.). So there is no difference on how it is done at any point.
Old 15th December 2017
  #510
Gear Maniac
 

Thanks pentagon.
How about when going from floating to fixed? Is it also trivial math?
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