How to choose between dithering plugins?

Hello folks,

This is sort of a technical question. The manual says that basically, you have to use dither for all projects, because as soon as you move one slider or insert one plugin, Wavelab applies 64bit precision.

So, being that you always have to apply Dithering, I would like to know:

  1. How do I choose which type of noise to use? (and please, don´t answer “whichever sounds best”. I´d like to know some specific, technical criteria)
  2. How do I choose which dithering plugin is best for the specific song or songs I´m mastering?
  3. Which is the best dithering configuration you recommend, and why?
  4. What are the risks of not applying dithering? How exactly will the quality of my mastered export suffer from not including dithering?

Please, be as thorough as you like. Links to articles or books are also welcome. :mrgreen: :mrgreen: :ugeek: :ugeek:

Happy weekend!

I like the iZotope MBit+ Dither that comes with WaveLab because of the auto-blanking feature, and the dither itself sounds good and works well.

For $19 USD you can get the excellent GoodHertz Good Dither plugin:
https://goodhertz.co/good-dither

It has a nice built-in bit-depth meter which I might have helped push them to add :slight_smile: But WaveLab also had a good/similar bit-depth meter built in now.

Also, saw this article today randomly that might explain things better:
http://audiohertz.com/2018/09/28/what-the-fck-is-dither/

Hi …

I think you could make the argument that a significant percentage of mastering engineers employ “flat” TPDF dither. WL built in MBit+ can be easily configured for that: Dither amount: High Noise Shaping: None.

Dither is likely to be the least significant factor in your track selling platinum/winning a Grammy or not.

Since you asked for book referrals, maybe take a read of The Principles of Digital Audio by Ken C. Pohlmann (he knows stuff). Very readable.

https://www.amazon.com/Principles-Digital-Audio-Sixth-Video/dp/0071663460

+1, if you want thorough, this is the book for you.

I hadn’t tried this in a long time and I still haven’t really tested with it, but it seems useful. It’s most noticeable in quiet parts of a track. https://steinberg.help/wavelab_pro/v9.5/en/wavelab/topics/master_section/dither_quality_test_c.html

If you bypass the dither plugin while in this mode, it’s often easy to hear the quantization noise.

The risk is It will sound like the test where you bypass the dither plugin in the previous example, although the quantization noise will be much lower level and much less apparent when it’s not being exaggerated by that tool. But it will still be there at a much lower level, which you can test by rendering to a 16 bit file without dither, and cranking up the very end of a fade-out in headphones. You might hear nothing terribly objectionable, or you might easily hear the quantization noise. But It’s really not safe to not use dither, especially when rendering to 16 bit.

Back to which dither to use, it’s my understanding that the safe dither to use when audio will be further processed (which is always the case when you’re talking about a wav file you’re making that will be converted to MP3, etc. for the stores and streaming) is TPDF with an amplitude of two quantization steps.
https://en.wikipedia.org/wiki/Dither#Different_types
(under “which types to use”)

Izotope has said that MBiT+, level high, no noise shaping is flat TPDF,

I’m not entirely sure about UV22.

I’m sure the Wavelab Internal Dither can be set to flat TPDF of that amplitude without noise shaping.

A user made a list of all the modes and specs of the included Wavelab dither options. If all of the information is correct, it would still apply to Wavelab 9.5.

https://www.steinberg.net/forums/viewtopic.php?t=85118

In the real world they may all be fine. I’m sure many commercial MP3s and AACs have been made from wav files that used dither with noise shaping, even though it’s not recommended.

Dither when reducing the bit-depth of digital audio is a significant part of the process which enables digital audio to work properly, both in theory and in practice. If you do not do it you are damaging your audio (and for no good reason, as the processing involved is trivial).

Paul

Thanks everybody! I didn’t think this post’d get the attention it did.

Keep the answer comming!

:smiley: