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
DAW manufacturers, please let us know where you dither! DAW Software
Old 14th February 2017
  #31
Lives for gear
 
stinkyfingers's Avatar
 

just a random thought on the test file...
does Logic still truncate 32 bit float files to 24 bit integer upon import ?
last i used was v9 and you could not really enjoy the full benefits of 32 bit float source audio files...
i would assume this would have an effect on the dither test ?
Old 14th February 2017
  #32
Lives for gear
 
ISedlacek's Avatar
The main question remains: do we really hear a difference in a real world on real music ? I tried long time ago and it was very difficult to blindly distinguish between tracks with different type of dither used and even between tracks when dither was or was not used at all (similarly as for example Ethan Winer demonstrated here few times).

After that I became much less worried about dither
Old 14th February 2017
  #33
Lives for gear
Not sure how honest they are but in Sonar you have the option of turning dithering on or off in the audio options section. They also give you a few choices of dithering upon export. With those up front options I kind of doubt there is any kind of hidden dithering going on.
Old 14th February 2017
  #34
Lives for gear
 
JP__'s Avatar
 

Quote:
Originally Posted by ISedlacek View Post
The main question remains: do we really hear a difference in a real world on real music ? I tried long time ago and it was very difficult to blindly distinguish between tracks with different type of dither used and even between tracks when dither was or was not used at all (similarly as for example Ethan Winer demonstrated here few times).

After that I became much less worried about dither
Interesting, I have completely different experiences in real world mastering scenarios (quite comfortable to check with RXs Dither-settings for example).
The practical use of realtime dither also depends on the used converters, processing and even the music material you are working with (sometimes rock music can be more sensitive than pure acoustic music...)
I even have done some tests for software/digital-audio-processing developers which where quite sceptical at the beginning. In my experiences its exactly like Bob mention here for years; its highly dependable and sometimes you can achieve quite big differences and sometimes its more or less not worth mentioning.
The conclusion for me therefore is: its just silly to not use any dither as the harm from truncation CAN be much higher than any added noise will ever be. Even when the diff is extremely small in a given processing stage (in most cases there will be additional processing added).
A pure theoretical discussion leads to nowhere here (as usual), its a practical topic.
Old 14th February 2017
  #35
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by JP__ View Post
A pure theoretical discussion leads to nowhere here (as usual), its a practical topic.
Oh, not sure about that, I have learned loads in this and the other thread.
Old 14th February 2017
  #36
Lives for gear
 
JP__'s Avatar
 

Me too, but in the end we have to test it ourself and see what setup works best in our case. Trust your ears and not the people who make smart sounding statements in a forum, either way.
Otherwise I see that a lot of people are overwhelmed which all those tiny differences and detailed decision to make (which makes fully sence especially when we talk about musicians who wants to concentrate on music not on technology) and therefore the good advice to better dither twice before not to dither. Theres no harm in dithering... But a lot of in theoretical discussion about hearing thresholds and distortion levels.

I mean two big threads about an topic no DAW user has to think about in a perfect world and at the same time so much ignorance from a lot of developers which prefer to juggle with numbers instead of being consequent. Its such a shame for "pro" audio...
Old 14th February 2017
  #37
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by JP__ View Post
Me too, but in the end we have to test it ourself and see what setup works best in our case. Trust your ears and not the people who make smart sounding statements in a forum, either way.
Otherwise I see that a lot of people are overwhelmed which all those tiny differences and detailed decision to make (which makes fully sence especially when we talk about musicians who wants to concentrate on music not on technology) and therefore the good advice to better dither twice before not to dither. Theres no harm in dithering... But a lot of in theoretical discussion about hearing thresholds and distortion levels.

I mean two big threads about an topic no DAW user has to think about in a perfect world and at the same time so much ignorance from a lot of developers which prefer to juggle with numbers instead of being consequent. Its such a shame for "pro" audio...
Well exactly.

And I have learned to dither before going out in any way and I also now do it when doing a bounce in place. And monitor through dither. As it seems in this less than perfect world where DAW companies don't do this for us en large it makes a useful difference. And that's it. These two threads have upgraded my knowledge just enough and my workflow in a simple, non obstructive way that makes a difference. Win.
Old 14th February 2017
  #38
Quote:
Originally Posted by ISedlacek View Post
The main question remains: do we really hear a difference in a real world on real music ?
The second half of the test file above is made to simulate a typical piece of music.

Generally, I thought that this thread was, refreshingly, not about "what you hear". Instead, the idea to ask DAW manufacturers what they really do, or in doubt, measure it for ourselves.

I think that it is beneficial for every audio engineer to know whether and when the application dithers automatically, "where is should". If it doesn't, you can safely expect other areas to be handled similarly (correct or messy).

Huge audio apps are typically made by relatively big teams in a corporate environment. This is really the last place to expect any member to find time for audio! Instead, daily work is about making coffee and how lunch will be. Then, blindly closing tickets as they pop up, fullfilling change requests in isolation and getting the next scrum goal done, hoping the master knows what he's doing. Real testing comes last, if there's time left. Corporate structures simply mean: Nobody has any relevant overview (they often have hyper detailed knowledge about their own expenses, because they depend on other ppl's money. But that's it!).

From my experience, small teams and one man operations deliver much better software from A to Z. That probably why corps like to buy them out.

Let's simply measure these apps and we'll see. Independently of what you can hear or not. Let's check which manufacturer really did his homework. I expect most to not fulfill the sampling theorem as they should (regarding truncation behavior). It would be a quite hefty surprise for software sold for several hundred bucks to show clear flaws in audio handling.

Last edited by FabienTDR; 14th February 2017 at 05:22 PM..
Old 14th February 2017
  #39
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by FabienTDR View Post

I think that it is beneficial for every audio engineer to know whether and when the application dithers automatically, "where is should". If it doesn't, you can safely expect other areas to be handled similarly (correct or messy).
My thoughts exactly. Only, this is clearly not the way to find out.
Old 14th February 2017
  #40
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by FabienTDR View Post

Let's simply measure these apps and we'll see. Independently of what you can hear or not. Let's check which manufacturer really did his homework. I expect most to not fulfill the sampling theorem as they should (regarding truncation behavior). That would be a quite hefty surprise for software sold for several hundred bucks showing clear flaws in audio handling.
Yup, same expectation here. Would also put a boot into the 'all DAWs are sonically exactly the same' sort of thing at least as a concept........
Old 14th February 2017
  #41
Quote:
Originally Posted by stinkyfingers View Post
i would assume this would have an effect on the dither test ?
It certainly would have. But doesn't it accept a 32bit floating point file? I thought that all modern audio apps will do, as they use this format internally anyway. Can't try here.

Not sure if it clear, but the test signal's format is 32bit floating point, not 32bit fixed. In doubt, I can create a 24 bit version.

Last edited by FabienTDR; 14th February 2017 at 05:42 PM..
Old 23rd February 2017
  #42
Lives for gear
 
stinkyfingers's Avatar
 

Quote:
Originally Posted by FabienTDR View Post
It would be a quite hefty surprise for software sold for several hundred bucks to show clear flaws in audio handling.
the sad thing is 11/10 people don't give a **** and they just want their trivial features rather than a "proper" DAW.
Logic has been "exposed" for (extremely) low quality/resolution automation/faders but nothing has changed. no one cares. they give you a cool virtual drummer with a hipster name and tattoo instead of addressing a flawed audio engine. thanks.
as soon as i found out how ****ty of a DAW Logic was, i dumped it and went to Reaper and never had any regrets.
16 bit dither aint gonna do **** when you have 7 bit fades.

im also a little surprised that Karloff70 did not come back to share Logic's results of this dither test. i can't imagine why...
Old 24th December 2018
  #43
Lives for gear
 
Robo's Avatar
I'll try testing some of the DAWs I have. I suspect Samplitude and SAWStudio should do well - both have various dither options, Samplitude supposedly works at 80 bit floating point, and SAW is fixed-point maths throughout.

I've found truncation distortion is quite easy to spot on reverb tails and fades, with the appropriate volume boost of course. With music it sounds a bit like a detuned radio, when the dial is not quite on the correct frequency. You can still hear the rhythm of the music but it's a crackly mess, and continues to break up as it fades out.
Old 8th January 2019
  #44
Glad to see this thread still alive. Dithering is quite a controversial topic apparently still in 2019, eh?

Those reading here make sure not to miss the "good dithering practices... what are yours?" amazing thread here at GS.
Old 8th January 2019
  #45
Lives for gear
 
Deckdaddy's Avatar
I think my reverb plugin sent to my mixing board sounds better when I dither after it last thing before the converters feed my mixing console.

I also keep a dither plugin in my monitoring chain in my DAW. I use NaturalizeDither by Airwindows, not a traditional dither I think as it’s not noise but some random Benford law statistic thingie.
Old 29th January 2019
  #46
Lives for gear
 
popmann's Avatar
Quote:
Originally Posted by stinkyfingers View Post
just a random thought on the test file...
does Logic still truncate 32 bit float files to 24 bit integer upon import ?
last i used was v9 and you could not really enjoy the full benefits of 32 bit float source audio files...
i would assume this would have an effect on the dither test ?
FWIW. It DOES reduce it. I've never examined how, but LPX will take in a 32float file and store it as a 24integer file on disk.
Old 29th January 2019
  #47
Gear Guru
I use HArrison Mixbus and not sure how the dithering is....... It does sound good as a DAW since it has a console sound built in
Old 29th January 2019
  #48
Gear Maniac
 
Sacha's Avatar
I want to test Cubase but I'm scared and also don't want to switch DAWs
Old 29th January 2019
  #49
Scared? This really goes too far. I'd suggest to find a problem first, then bother.

I'm a greatly into doing things right. But doing so without being able to see or experience any issue, is straight paranoid, likely harmful at the end of the day.


Take note how little ABX tests or serious measurements you find in these threads. Take note how little reasonably reproducible and verifiable theory you see in these threads. Take note of our very well established thresholds of perception and how distant these digital ghosts are from these. Take note how easy it is for our perception to imagine things, especially when there is nothing to find, or worse, when we start to expect things.

The more irrelevant the matter, the more vivid the public debates. This vividness is not to be confused with relevance. Google "bike shedding".

The sampling theorem proves its completeness (that's why it is a theorem). Digital capture, synthesis, processing and reproduction of continuous signals hasn't been "found" by try and error. The system is proven from A to Z, and works so well for so cheap, for so long, for a reason!

Dither is a trivial part of it, with audio signals (very complex, noisy signals) anyway. Modern hi resolutions push the matter into complete irrelevance (both theoretically and practically/measurably).

Don't worry, this is will neither make nor prevent a great mix.

Last edited by FabienTDR; 29th January 2019 at 11:59 PM..
Old 30th January 2019
  #50
Gear Maniac
 
Sacha's Avatar
Quote:
Originally Posted by FabienTDR View Post
Scared? This really goes too far. I'd suggest to find a problem first, then bother.

I'm a greatly into doing things right. But doing so without being able to see or experience any issue, is straight paranoid, likely harmful at the end of the day.


Take note how little ABX tests or serious measurements you find in these threads. Take note how little reasonably reproducible and verifiable theory you see in these threads. Take note of our very well established thresholds of perception and how distant these digital ghosts are from these. Take note how easy it is for our perception to imagine things, especially when there is nothing to find, or worse, when we start to expect things.

The more irrelevant the matter, the more vivid the public debates. This vividness is not to be confused with relevance. Google "bike shedding".

The sampling theorem proves its completeness (that's why it is a theorem). Digital capture, synthesis, processing and reproduction of continuous signals hasn't been "found" by try and error. The system is proven from A to Z, and works so well for so cheap, for so long, for a reason!

Dither is a trivial part of it, with audio signals (very complex, noisy signals) anyway. Modern hi resolutions push the matter into complete irrelevance (both theoretically and practically/measurably).

Don't worry, this is will neither make nor prevent a great mix.
I love you
Old 30th January 2019
  #51
Lives for gear
 

Hi,FabienTDR.
Please make VST3 version of Nova GE for Mac with 64 bit processing.
I can not use VST2 in Cubase,its only 32 bit
Old 30th January 2019
  #52
Gear Maniac
Ableton Live has a Audio Fact Sheet in the manual:

Audio Fact Sheet — Ableton Reference Manual Version 10
| Ableton


It lists "Dithering" as Non-Neutral, implying all operation listed as Neutral don't dither.

The sentence "Dither should never be applied more than once to any given audio file, so unless you are mastering and finalizing in Live, it is best to always render at 32-bit and avoid dithering altogether." together with "Live‘s internal signal processing is all 32-bit" implies no dithering at all in Live's internal processing, only as an option when exporting at lower than 32-bit.
Old 8th February 2019
  #53
Gear Nut
 

Time to report
Old 8th February 2019
  #54
Lives for gear
 
stinkyfingers's Avatar
 

Quote:
Originally Posted by popmann View Post
FWIW. It DOES reduce it. I've never examined how, but LPX will take in a 32float file and store it as a 24integer file on disk.
And it does not dither, right ? (Didn’t in 9)
Old 8th February 2019
  #55
Lives for gear
 
popmann's Avatar
Quote:
Originally Posted by stinkyfingers View Post
And it does not dither, right ? (Didn’t in 9)
I have no idea.never imported dead air to test. But to be fair, not applying any DSP/volume change, you should be able to pull the integer back out without. Dither becomes needed if they were also doing some DSP on the import. Me thinks. Me is wrong a bunch.
Old 8th February 2019
  #56
Quote:
Originally Posted by goshalev View Post
Hi,FabienTDR.
Please make VST3 version of Nova GE for Mac with 64 bit processing.
I can not use VST2 in Cubase,its only 32 bit
VST2 generally supports 64bit I/Os (as part of an extension developed by Steinberg themselves). All our recently updated product also support 64bit I/O. Please blame the host for not doing so (or even building back support)!

Reaper has no issue with 64bit I/O, neither has it sidechaining issues with VST2. Worth a try, it also offers Cubase skins.

VST3, maybe at some point (maybe end of this year). But to be clear, it's pure additional costs, without giving us any advantage. The restrictions you mention are artificial bottlenecks imposed by Steinberg only (and Steinberg hosts only). They could change their mind anytime and do the same again once they come out with VST4, e.g. by blocking stereo processing for VST3, or forcing VST3 UIs then to appear black and white. Not my world. I don't want to support this crappy behavior.

Motivation on our side isn't exactly glowing for VST3 to say the least.


- Sorry for the OT -

Last edited by FabienTDR; 8th February 2019 at 08:55 PM..
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