I appreciate your effort in conducting these tests, but I am particularly interested in knowing if you were able to perceive the differences in sound and how you would describe them.

 

Why? What then?
It mostly is searching for approval.

Someone I rely on made a beautiful statement some time ago, when someone asked for an eq review. "You really want me to for example raise 8kHz and tell you that there are more highs then?"

Recently I saw a conversation between Sloterdijk and Žižek. Sloterdijk made a very interesting statement there. "The biggest enemy of evidence is not the lie. The biggest enemy of evidence is perception."

Yeah, if one checks the foundation of statements that clearly ignore evidence, the argument is indeed always perception.


So regarding the statement you made some time ago about you advising to rather hear then measure. In a technical field, objectively it really is not. And audio processing is extremely technical.
A nulltest is enough for an objective statement about the difference between two or more audio signals.

Everyone in the audio processing field calls himself an "engineer", but everyone is talking about audio like Marry Poppins would talk in no mans land.

 

Last edited: Jan 3, 2024

Bob Katz once said about dithering: If you are in doubt, don't dither at all. I say about dithering: I don't care, I don't dither. I never heard a problem nor did I felt a problem. So I am with Katz and Sloterdijk and without dither. And while using exclusively 32bit floating WavPack-files there is noone in this world - no human - who can hear dither or no dither at -750 or so db ... and I don't care what nulltests in that area say.

 

Most are talking about the science itself but completely avoiding the topic of how different dithers SOUND. In my experience, there is absolutely a different sound in different dithering algorithms. I may be an edge case as my hearing is hyper sensitive (consistently hearing differences as small as 0.01 dB) so YMMV. Not trying to start arguments here, just sharing my hopefully useful opinion. Especially since I have 1500 plugins installed and have tried all of them.

*INHALES*

First off, before introducing recommendations for plugins and subjective opinions, we need to explore what the fuck dithering actually is. Am I gonna be the one to explain it to you? NOPE. Instead, how about a 40 page PDF directly from iZotope on the topic?

Now, assuming you survived the driest (yet interesting) PDF you have ever laid eyes on, I'll move onto my own opinions.

If you want an IMO obvious difference in dithering sound, I suggest trying my absolute favorite (strictly) dithering plugin A.O.M. Sakura Dither. There are 4 different algos in this plugin and they all sound noticeably different.

• Type 1: Natural / Transparent
• Type 2: Open / Wide
• Type 3: Warm / Slight de-esser feelings
• Type 4: Electronic / Enhancement

Type 4 is my favorite most times. Really does something sweet to low end / transients. These types really give you a final control over a flavor of your tracks and I don't believe there is a better dithering product out there.



As a runner up, we have Goodhertz Good Dither. While less "boutique" sounding than Sakura, these dithering algos feel much cleaner, clearer, and more agreeable to my ears than something like Pro-L2 or other traditional types such as POW-R. Goodhertz provides a couple of helpful visuals on their website for comparison as well.

My favorite settings are: Optimal Dither Amount and either Optimal or Flat Noise Shaping.







In third place, I gotta give it to granddaddy iZotope. Don't think I really need to provide a screenshot for this one lol. Over the years (since the early 2010's) I'd guess that their algos were the most used (and misused) out of any plugin. To my ears, they have always been confusing. I couldn't really tell what they were doing or if it made my results better or worse. My inclusion of iZotope on this list is only because I know it's so widespread in its use. But I NEVER use it. I'd rather use Sakura, Good Dither, orrrrrrrrrrrrr........... my next recommendation.

MAAT FiDef is more psychoacoustic mumbo jumbo than dithering. Yet, the concept is similar to dithering with noise and shaping and etc. Except the website states it is DEFINITELY NOT DITHERING. It does something more subtle than Sakura, yet also completely unique compared to other options. Out of all 3 of the other options, this one may be the most difficult to hear. My favorite setting is EVOSE, which according to the manual is good for smiley face shaped mixes.



If I only had to pick 2 plugins out of all of these, it's Sakura and FiDef (in that order). I also combine both of these sometimes but you can tell when you stack them it can become a little bit too much added. Though, when they work together it's very cool. I'd ususally go FiDef first, then Sakura second.

Any questions, concerns, or opinions are welcome. Happy new year!

 

Last edited: Jan 3, 2024

Tell me you can' hear heavy preringing without telling me you can't hear heavy preringing: I USE FIDEF!!!

 

The topic is how dither affects sound when placed at different places in the chain, particularly in the files pratyahara provided (and it's possible uses as a crative tool if i understood correctly?).

 

True, the topic is about dithering in different places in the chain. But I think there's a bigger point to be made about the choice of NOISE TYPE in respect to different places in the chain as well. I think that yes, dither (and other noise) can be used as a creative tool. On busses and individual tracks to give a specific flavor (especially in reference to Sakura above). But, following the PDF from iZotope about when and where to dither applies only to a final product. I am firmly in the camp of finalizing with dither. If you take and put it first in the chain, being followed by saturation, EQ, compression, etc then the exact shape and amplitude of the dither is compromised and does not set in the ideal way it is supposed to. By that I mean ideal as in the way the dither is tuned by the plugin company or specific algos like POW-R.

Dither is different than taking the V-Gain in a BX channel strip (for example) and pushing it up. This is indeed also a type of noise, but still different than dithering. Which is why having X amount of channels of BX channel strip with noise in each of them is fine, but applying DITHER in the mastering / finalizing stage is a different beast. If we step out into analog where each piece of equipment has a noise floor (especially tape machines) then cumulative noise is only an issue if you don't account for it. Usually in the recording process (loud enough to escape the floor but quiet enough to avoid clipping).

But in the DIGITAL world, noise is a different beast. Take Waves (ew) "analog" buttons for example. Literal trash just pumping in nasty noise. But compare one of those buttons to a tailor made DITHER, and you can immediately tell it's different. Not just in shaping but in specific level as well. So, in sum, dither as a CREATIVE effect wherever in your project is absolutely fine. If it gets you to a result you enjoy more, go for it! But if we are talking about the "ideal" dither sound, it is absolutely LAST in the chain because that is what every single dither algo that exists is designed for.

Then again, all that matters is the results to your own ears. Do what sounds good!

 

My personal opinion about this and most other threads by pratyahara about cpu cores, os services optimization or whatever for "better" audio quality is just like religion. He/she can't prove anything, but if it makes him/her happy, who cares...

And this whole "you can barely measure it, but I can clearly hear a difference at -125 dBFS" thing is just soo plain ridiculous. Every breath or blood pressure change we constantly have makes a big more difference in our hearings noise.

But people kill each other because of religion for thousands of years, so this is most likely just human.

 

Dan Worral did dither episode and what i got from it was: As long as its not reapers dither, you are on the safe side

 

~ 120dB below peak? How could I - or anyone else? Honestly, I've made various double blind tests, several times, there's no difference in sound, transients or whatever. Everything that remains in a phase inversion test is a constant white noise with a bump above 18kHz. The perceived differences are just psycho acoustics. I recommend to try the following: listen to one 2-sec sniplet and try to recognize it in a double blind test among the three. You can't. You can't even if you compare it to only one of the others. Of course you need to make this test 10 times or even more.

 

So I can say the same about Audacity. It was old version. I hope it was only temporary bug and I was unlucky but dither in Audacity gived me huge artifacts on bass sample.

 

Some people place greater faith in formulas than in themselves. If a formula fails to bring them joy, they are unable to feel it. However, they find satisfaction in the adherence to the formula.

 

Thank you for your sincere response. However, based on your reasoning, it might appear that your inability to perceive any difference is not due to the technical facts you've mentioned. It's possible that these facts instead create a 'technical/scientific' placebo effect, making the difference 'impossible to perceive.' If this isn't the case, I can cite numerous instances where individuals, including renowned music professionals, couldn't distinguish between mp3@320 and the original WAV formats. People's hearing abilities vary widely, making such discrepancies quite common and understandable.

 

Apogee uv22hr on spectrum?

 

Perception plays a pivotal role in acquiring evidence; we wouldn't even recognize something as evidence (as anything, in fact) if it didn't register with our senses. Any scientific verification ultimately culminates in the perception of facts, often referred to as experimental proof.

 

If you detest ad hominem arguments, you wouldn't write things like this. Labeling someone as a fantasist or a fanatic and encouraging others to ignore them is a cruel form of such argumentation.
Gradually, this discussion shifted towards testing hearing abilities. This was not my wish. The case of Bob Katz illustrates how even a renowned audio designer may struggle to discern the distinction between a dithered and a non-dithered file. Many individuals who can perceive the difference feel empathy for him, which is understandable. Despite his ability to conduct the most intricate technical tests imaginable, this natural limitation persists.

 

Last edited by a moderator: Jan 4, 2024

I suppose so, yes.

 

Confirmed.

 

It's strange that no one asked what this beautiful piece of music was. It's Vivaldi's 'Orlando Furioso' RV 728 - Ritornello.
Version "after" is properly dithered.

 

Last edited by a moderator: Jan 5, 2024

I dither about finishing a mix, not dithering a mix, to be honest with you.