how does C6 handle internal bit resolution vs. exported res

this is a question that relates to internal bit length handling and what happens on export.
example:
24bit tracking>
put a bunch of plugins for processing (i.e. UA Massenburg EQ+Fairchild compression+Limiter) across the master bus>
mix down master bus to 32bit floating file>
analyze in WL>
result: “apparent” bit resolution i.e. 44bit (or higher/lower, but more than 32bit)

Q: if i wanted to save the mixdown file in 24bit format, would i need to dither down to 24bit?
or does Cubase internally process the mixdown without truncating the word length to 24bit, in which case dithering would NOT be necessary?

thanks

Technically yes you should use dither when going down bit depth.
In wavelab you can use the built in dither and switch off “noise shaping” which is relatively important if you know further processing and dithering will be applied later on. Noise shaped dither should only ever be applied once and be the very last process.

The strict technical answer may be yes, but you will NEVER hear the difference going to 24 bit. The will be an audible difference when going to 16 bit, but even that is not very noticeable in many cases.

FWIW - I never bother to dither when going to 24 bit, but I do for 16 bit exports. There is certainly nothing wrong with adding the proper dither when exporting a 24 bit file, but I consider it a waste of time.

Hi. :slight_smile:

If you can hear the difference between a 24 bit file and a 16 bit file that has been properly dithered from 24, I’ll eat my shoes. Both of them. :sunglasses:

I was talking about the difference between a 16 bit file that has been truncated from a 24 bit file versus a properly dithered 16 bit version of the same 24 bit file. It may take turning up the volume during a reverb tail to hear the difference, but it can be beneficial to use dither when exporting to 16 bit. I believe the importance of dither is often overstated and becomes ridiculous when going from 32/float to 24 bit, but there are purists who insist it must be done even though it will never be audible to any human. If I forgot to add dither to a 16 bit export I would not necessarily loose sleep over it, but it is good practice - dithering a 24 bit export is a complete waste of time IMHO.

besides the question if things are “audible” or not etc. I am interested how C6 handles the generated word length.
internally everything is apparently run in 32bit floating point, which makes total sense considering the enormous amount of processing by many plugins and many tracks recorded at least at 24bit and given the fact that even changing only the volume of a 24bit track/file generates a greater word length than the original 24bit.
so anything at the output of the master bus has more that 24bit. so it should be absolutely necessary to either mix to a 32bit floating point file format (which can not be handled by many non Steinberg DAWs or mastering programs) or it must require dithering applied if going down to 24bit. if that is the case then the final CD master would factually have dither applied twice. once from C6 to 24bit and in the end to bring the whole CD master to 16bit. Not good i think.
if i want to submit 24bit mixes for mastering where they can not handle 32bit float files, how can i avoid this double dithering?
do i understand split’s post right that he suggest to switch off the noise shaping, thus achieving a bit reduction without truncating and without adding any additional noise?

thanks

i hear fishtanks argument. although “waste of time” does not really come into play. you pop the dithering plug in that last master slot and the mixdown takes maybe 5sec longer…
the question is if there is an audible difference between a double dithered track (C6 to 24bit, then CD master to 16bit) and one that has only the final dither applied in the last mastering stage.

The point being made is that dithering from 32bit FP to 24 bit, the rounding errors are so low that it’s inaudible. I stated that “technically” dither should be applied whenever a files bit depth is reduced. I also stated that noise shaped dither should only be used once and be the very last process applied.

Any dither noise applied to a 24 bit file will also be inaudible when converted to a 16 bit file and final 16bit shaped dither is applied.

Non noise shaped dither still uses random noise to hide the rounding errors.

Cubase does not automatically apply dither when exporting, it also does not have a plugin for Non Noise shaped dither.
So you either use Wavelab to Dither from 32FP to 24 (non noise shaped) or don’t apply any dither 32fp to 24. use your ears to help you decide if you think dithering is necessary.

Like Split says, dithering from 32f to 24 is virtually inaudible. Another point: any mastering facility not able to handle 32 bit float files is suspicious IMO. I can’t think of any reason a masterer could not handle them.