Jump to content

Dither ?


Wolffman
 Share

Recommended Posts

I've been reading the thread double dither in the main forum and as a result I have realized I have been doing just that on some occasions with my own recordings. I'm a V8 SE user and the options for NOT dithering are.

1: No Dithering, 16 Bit internal resolution. ( not recommended for volume,effect or mix calculations )

2: No dithering, math rounding of sample value

So my My question is, if i have a limiting plugin on the master out that has dithering and i,am mixing down to a 16 bit, 44.1 stereo file which one of the no dithering options is the best one to use?

or is it better to turn of the dither function in the plugin ( if possible ) and use sams ( Dither with triangular spread noise )

i beleive this is the best one to use in V8 SE as it does not have the POW-R option of the full versions.

Thanks in advance for any help ;)

Link to comment
Share on other sites

I would follow the advice "not recommended for volume,effect or mix calculations" - if you use this mode,

everything you do within your project is done in 16 Bit instead of 32.

So I would either use no dither/math rounding + the plug-ins dithering, or turn the plug-in dither off and use

samps dither/triangular spread noise. For this decision you have to listen which one is better for your material.

Regards

Ben

Link to comment
Share on other sites

Thanks Ben

I'll try your sugestions and see if i can hear the difference, from what i've been reading the noise dithering adds is very low level. Are there any simple methods to compare one type of dither against another?

Maybe theres a thread on this subject ?, if so could someone steer me in the right direction

thanks again. ;)

Link to comment
Share on other sites

It's important to understand that dither only adds enough noise to eliminate the distortion. If you don't dither, the distortion can be considerably more audible and will build up much faster upon subsequent processing than the noise level from dithering each bit reduction. Dither doesn't just cover up distortion, it prevents it.

I look at it as being a lot like the use of high frequency bias in an analog tape machine. Yes, you can record with it turned off but why do that?

Link to comment
Share on other sites

Thanks Bob

So basicly its better to dither a couple of times if necessary than not to dither as this will stop build up of distortion ? And the less dithering the better i guess.

If i have a multi track project i want to bounce to a stereo 32 bit float file for further processing does samp dither this file or not? Do i have to switch dither of or should i leave it on?

( is it necessary to dither this 32 bit file or not ?)

Sorry about all the questions but i'm just trying to understand this whole dithering concept as it is obviously more important than i originaly thought in getting the best quality recordings possible, ( our common goal ).

Regards

Wolffman

Link to comment
Share on other sites

If i have a multi track project i want to bounce to a stereo 32 bit float file for further processing does samp dither this file or not? Do i have to switch dither of or should i leave it on?( is it necessary to dither this 32 bit file or not ?)
When bouncing to 32 bit float SAMP doesn't dither (because dither isn't needed in this case), so you don't need to switch it off.

All integer bounces (24, 16, 8 bit) are dithered with the right amount, this means if you bounce to 24 bit the dither level is about -140 dBfs. Therefore in 24 bit bounce you should leave it on and don't need to worry about double dither as the very low amount of dither doesn't hurt. (the best DA-converters can do about 120 dB dynamic, for comparison).

Bouncing to 16 bit you need to decide - original 16 bit source files WITH NO CHANGES AT ALL do not need dithering - there is no information that can be recovered by it.

All others situations need dither.

Regards

Kai Schluenz

KATAPULT Tonstudio

Link to comment
Share on other sites

Bouncing to 16 bit you need to decide - original 16 bit source files WITH NO CHANGES AT ALL do not need dithering - there is no information that can be recovered by it.

these 16 bit source files with NO CHANGES AT ALL? do you mean recorded tracks with no change to level or compresion,reverb etc applied? or something different.

Link to comment
Share on other sites

Bouncing to 16 bit you need to decide - original 16 bit source files WITH NO CHANGES AT ALL do not need dithering - there is no information that can be recovered by it.
these 16 bit source files with NO CHANGES AT ALL? do you mean recorded tracks with no change to level or compresion,reverb etc applied? or something different.

NO CHANGES AT ALL is exactly describing the situation. ANY CHANGE, be it as small as you want, requires redithering to gain optimum results as it produces information below the 16 bit range. This would be lost if no dither is used.

To be correct: an exaxt boost of 6.020559991328 dB or a multiple of this number would not require dither, as it's an exact shift of one bit up (level x 2). But you cannot set this exact number, so it's only theoretical.

On the other hand - if you burn a 16 bit file to CD without changing anything, e.g. a copy of a CD or 16 bit recording, you don't want dither to be applied, as this only would add (dither-) noise and wouldn't recover any information below 16 bit, because there is non.

You can watch the bit activity display in the visualisation if you want to be sure - if it shows 16 bit on the mix, don't dither.

Regards

Kai Schluenz

KATAPULT Tonstudio

Link to comment
Share on other sites

Introducing noise can make low-level information more audible but this has nothing to do with dithering.

A definition of dithering that I like is redistributing bit reduction artifacts so that they sound like noise rather than like distortion. I do a demonstration for my students by setting up a noise gate threshold so that it chatters so bad that the music is hard to hear. Then I mix some noise in and demonstrate how eqing the noise can make the sound amazingly listenable.

Link to comment
Share on other sites

Introducing noise can make low-level information more audible but this has nothing to do with dithering.
This exactly IS dithering: adding noise to keep the LSB active and lift information below the quantisation resolution above the LSB threshold.
A definition of dithering that I like is redistributing bit reduction artifacts so that they sound like noise rather than like distortion.
What you're talking about is noiseshaping - compensate low frequency quantisation artifacts by less- or inaudible high frequeny signals generated from the audio AFTER quantisation.

This does NOT recover information that wasn't detected by the AD-converter.

In praxi dither and noiseshaping are often combined to "dithered noiseshaping".

This may not be mixed with "noise shaped dither" wich is dither with a special equalization pronouncing the less audible freuencies above 10 kHz.

I do a demonstration for my students by setting up a noise gate threshold so that it chatters so bad that the music is hard to hear. Then I mix some noise in and demonstrate how eqing the noise can make the sound amazingly listenable.
Then you're showing your students a misguiding simplification, as a noisegate incorporates time-constants.

If it hadn't those it just would be a nonlinear transfer funktion causing the usually called "crossover distortions" and wouldn't be called noisegate.

Digital quantization could be viewed as a couple of those nonliniarities, build by the quantization steps ("bits").

It would be better to show the student a system with low bitrate, e.g. 4 or 8 bits and a high samplerate, 48 or better 96 kHz.

You can use a 16bit converter and boost the outpult in the digital domain by 64 or 48 dB if you don't have a 4 or 8 bit AD.

Then, in the first step, add audible white noise, and then inaudible noise (20kHz-fs/2) to show the effect.

Finally, if available, you could show noiseshaping, where the quantasation noise is shifted to inaudible frequencies.

The (Mac OS9) freeware PROSONIQ SoniqWorx is very good for this, it has all the tools, including noiseshaping WITHOUT dither.

You can even "simulate" the AD converter in the digital domain here by truncating the lower bits.

Regards

Kai Schluenz

KATAPULT Tonstudio

Link to comment
Share on other sites

You can watch the bit activity display in the visualisation if you want to be sure - if it shows 16 bit on the mix, don't dither.

Can't seem to find this feature in V8 se, maybe its only in the pro version of samp?...

I don't use V8_SE, so it might be that it's not in there. Then you have to decide on the other criteria I gave you.

Regards

Kai Schluenz

KATAPULT Tonstudio

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

×
×
  • Create New...