why is dithering so important in mastering!

[sadboy]

guys, whats the effective importance of dithering in mastering... i know its important but what would it add to my master ?

[Adam Dempsey]

Short answer: minimizing your losses, retaining resolution of a higher bit depth/wordlength source.

Long answers (via the Search function, above):
To Dither or not to Deither?
Why Not Dither for Mix-Down?
Dithering down to burn audio cd in Logic
Izotope RX Resampling - Dither Necessary? (32-bit to 24-bit issue)
dithering needed?

[heybub419]

If you don't dither while converting wordlengths, your tracks will sound "buzzy" (for lack of a better word) at reverb tails and fades. You lose depth in the recording. Dithering helps this by sort of covering up this buzziness; by applying a noise signal that smooths out the least significant bit (last bit in the digital word).

[CJ Mastering]

i know its important but what would it add to my master ?

You dont really wan it to add anything noticeable

Dither adds low level noise to mask and to get rid of any quantization errors (mathematical errors) when going form a higher to lower bit depth
Cj

[sadboy]

thanks guys, really appreciate it.....

[dcollins]

After all this time, why do people still say that d*ther 'masks' or 'covers up' distortions?

Strange.


DC

[Waltz Mastering]

I have a feeling the ghost of Oky has just been summoned.

[jackthebear]

Originally Posted by dcollins

After all this time, why do people still say that d*ther 'masks' or 'covers up' distortions?

Strange.


DC

It's everywhere on the internots.....so it must be true.....

I get asked a lot about it......what does Ted Jensen use???

[jayfrigo]

Originally Posted by dcollins

After all this time, why do people still say that d*ther 'masks' or 'covers up' distortions?

Strange.


DC

Same reason they still believe there are "stair steps" on the output of their DAC.

Baffling. It's not like this stuff is new or not well-understood by science.

[dcollins]

Originally Posted by Waltz Mastering;

I have a feeling the ghost of Oky has just been summoned.

He comes back every couple of weeks under a new login name. Easy to spot even when the topic isn't d*ther.


DC

[Jon Hodgson]

Originally Posted by sadboy

guys, whats the effective importance of dithering in mastering... i know its important but what would it add to my master ?

It's more what it doesn't add when you requantize to a lower word length..

Quantization distortion.

You exchange it for a bit of noise, white noise in the simplest case (already pretty unobtrusive at 16 bits), spectrally shaped to be less detectable (hopefully) in the more advanced cases.

[Washington]

Just for the next fella who'll care to use the search engine on that subject... Here's a quick quote the best explanation I found on dithering, for which the guys at Izotope should be granted a Nobel Prize of vulgarization (the complete explanation, longer, can be found on their website) :

Consider your 24 (or 32, or 64) bit file to be an original photography, with all shades of grey from black to white. If you need to print it at a large scale, you'll have only black ink and white paper at your disposal. No grey at all : that's how printing works. THAT would be your 16-bit result. You've reduced the means at hand, but still want to convey the same image.

The function of dither is to act like a screen acts in printing. Adding a very low layer of shaped noise is comparable to "dividing" your image in an array of very small ink dots, small enough (inaudible enough in the case of the shaped noise) to trick your brain in believing you're seeing a shade of grey. Or to reveal nuances that would otherwise be lost in the wordlength reduction.

That's where the comparison ends. Dont' use a screen while printing and you'll have a result resembling the famous portray of Che Guevara we all know, made only of solid black and white patches with no in-between grey. Don't use dither when reducing wordlength and you'll abruptly cut the lowest (in intensity) signals to silence (your reverb tails, particulalrly, will be abruptly muted).

[heybub419]

I'm sorry if I was incorrect before. I've still got a LOT to learn, and I realize that there are alot of professionals on here who have much, much more knowledge and experience than I do. I was just giving an example I guess, of something I noticed before with dithering (the tail on a mix). I shouldve just held my tongue and read what a pro said! ()

[heybub419]

^ I read in Bob Katz's book about fading with dither vs fading with truncation, and so I tested it on a friends mix. It was very subtle, but when listening closely I noticed that with dithering, it seemed to extend the fadeout. This could just be a placebo effect...like something I wanted to hear? I'm still learning though, and thanks for pointing out that I was wrong

[Washington]

Good God, I didn't intend to say you were right or wrong. I just meant to add a bit of info, as dithering is not the easiest subject to grasp (wait till we enter the lovely world of jitter, though). Besides, I don't find the Bob Katz's Bible - which is however very complete and in-depth - to be so easy to understand at beginner/intermediate level.

If you are interested, download the dithering manual AND the sound files in this page (even if you don't have Ozone, it doesn't matter at all), and read carefully the instruction before listening (you'll need to crank your monitors at a dangerous level to hear things clearly). Real eye-opener on the effect (and correct use) of the dithering process.

[Jon Hodgson]

Originally Posted by heybub419

I'm sorry if I was incorrect before. I've still got a LOT to learn, and I realize that there are alot of professionals on here who have much, much more knowledge and experience than I do. I was just giving an example I guess, of something I noticed before with dithering (the tail on a mix). I shouldve just held my tongue and read what a pro said! ()

Your observation was fine, it was more your analysis, or at least your choice of words in presenting it, that was flawed.

Dither doesn't "cover up" the quantization error, it changes its nature from being signal dependent (distortion) to random (noise). The point at which this will be most noticeable will be when the difference between noise and distortion is greatest, which means small simple signals, like the tail of your reverb.

The quantization distortion isn't covered up by the dither, it's gone, in its place you have noise.

[heybub419]

Thanks for the link!

Sorry if I misread your post, I also meant to say that to the others on the thread, as well.

[heybub419]

Originally Posted by Jon Hodgson

Your observation was fine, it was more your analysis, or at least your choice of words in presenting it, that was flawed.

Dither doesn't "cover up" the quantization error, it changes its nature from being signal dependent (distortion) to random (noise). The point at which this will be most noticeable will be when the difference between noise and distortion is greatest, which means small simple signals, like the tail of your reverb.

The quantization distortion isn't covered up by the dither, it's gone, in its place you have noise.

Okay thank you... I did choose the wrong words to describe it. I know that dithering changes the nature of quantization noise as you said, but I should have picked a better way to describe it. Also, I was trying to think of an example of when you might notice it's effect on a signal...that's why I chose the "buzzy" example lol

[jayfrigo]

No worries. Thanks for taking part in the discussion just the same.

[UnderTow]

Originally Posted by dcollins

After all this time, why do people still say that d*ther 'masks' or 'covers up' distortions?

Strange.


DC

Some people never learn...

Alistair

[Babaluma]

Originally Posted by UnderTow

Some people never learn...

Alistair

well what's a better explanation, in your esteemed estimation?

[UnderTow]

Originally Posted by darkflame

well what's a better explanation, in your esteemed estimation?

Linearisation of the quantae. Not an explanation but rather a description of what it does.

Feel free to use the search feature. This topic has been done ad nauseum.

Alistair

[Dean Roddey]

Dither is important on Gearslutz because, otherwise, after we got done arguing about the 50 to 100 other things that are vastly more important to music making, we'd have to stop arguing. And that just can't be allowed. Dither has the lovely benefit of being one of those angels on the head of a pin type subjects that can go on forever, so it serves a very important purpose here.

[Dean Roddey]

Hmmm... Some of that sounds familiar. There used to be a guy around here who, in any discussion on dither, would go ballistic with various types of arguments of that nature.

I'm not sure I even understand what he's trying to say, but he seems to be skipping the word length truncation step, without which there's not really any need for dither anyway.

[UnderTow]

Originally Posted by the gaze

Actually, that's not what it does.

I'm glad you argumented your point so well or you might have looked a real pillock.

Did you mean to say "linearization of the quantizer"?

No, I'm not American.

Funny, but many who are posting with such smugness and "authoritah" don't seem to really know how dither works, or what it does [or even how to spell it!!!]

As opposed to you who write with smugness and say absolutely nothing?

So Mr Jones, will you please enlighten us with your wisdom and share with us what you think dither does? I'm up for a giggle right about now.

Alistair

[Jon Hodgson]

Originally Posted by the gaze

And while dither will literally get rid of quantization noise completely and only increase the noise floor by 3dB, dither will do nothing to get rid of the analog noise recorded from your mic preamp.

All fine and dandy if you're just going to record a 24 bit signal from your preamp and then ship that. You'll have quantization noise (not distortion, the signal is random enough to make the quantization random, with a pretty well even spectrum distribution), and it'll be at a much lower level than the noise from the preamp anyway.

But truncate that signal to 16 bits, and you're probably looking at quantization steps which match or exceed your original noise level. Most of the time you probably won't notice (as I mentioned before, dither is most needed when the signal is small and simple), but when you get to where it's fading out, then you may well do.