Dithering (Cubase)

davecg321

New member
Most of the time I master my tracks on the stereo bus (My projects are recorded in 24 bit) placing the UV22 plugin at the end (set to 16bits hi)

Basically I want to know if this is ok....?

or, should I be doing something else?

much appreciated

D
 
For something like this, there is no right or wrong way. There are more common and less common ways. Most people will export their mixes to a stereo wav file without anything on the master buss. Then either bring it back into their DAW program as a new project or load it into a program for mastering; like Wavelab. Dithering is applied during the last convert-down step. 32bit or 24bit to 16bit.

But if it is working for you on the master buss, then it should be fine. Dithering at the end of it all.

hope this helps.
 
It depends on what you are doing. Like Chili said, most people export the mix as a 24 bit wav file without anything on the bus. This does two things for you:

1. it gives you a high quality non mastered version of the raw mix. This is important if you ever compile these songs and send them to a mastering engineer.

2.It future proofs the mix. As time goes on and cubase keeps updating, some of your plugins will not be available, the session might not open, etc... If you needed to send the mixes out for mastering, you would have to start over again and remix everything.

BTW, what is 16 bit hi? Are you mixing down to mp3? If that is the case, you will want to make a copy of the mix without any mastering as a 24 bit wav file. Just like anything else, there are trends in mastering. 5 years from now, you may regret only having a low bit, processed copy.
 
Dithering/truncation should always be the last process applied. My question is this: if you record at a sample rate other than 44.1, does the DAW apply the sample rate conversion or the dither first when you export the mix? I would assume it applies what's on the master bus first then converts to the final format. If that's true then the "correct" thing to do is export the mix at 44.1/24 and dither/truncate as a separate process. If you record at 44.1 you could probably skip this extra step and export directly to 16 bit.

But I agree with the above responses, it's best to keep a high quality 24 bit "pre-master" at the same sample rate as the projects. If you record and mix at 96/24, export the pre-master at that setting with no processing on the master bus.
 
Last edited:
Another vote for exporting a 24bit mix and then master, for all the reasons mentioned above. I recently redid my mastering process and because I had raw 24bit mixes for all my most recent but still unreleased works, I was able to rather quickly remaster them all in an afternoon using the improved process. Flexibility and extra copies of things is never bad considering how quickly this field and technology evolves.
 
Dither should be applied before any process that will requantize the audio. A lot of it is source dependant as well, but if you do a 24 bit "safety" mix, it wouldn't hurt to apply dither. You'd be using it again to master to whatever end format you want. Any distortion you might get will accumulate faster than the noise added from dither. Dithering when you render the mix is your only chance to remove the distortion. As a final step when mastering, you can dither and apply noise shaping to move the noise away from the midrange where it's easiest to notice. You should only apply noise shaping once, so that would be when the target files will not undergo any further processing.
 
Dither should be applied before any process that will requantize the audio. A lot of it is source dependant as well, but if you do a 24 bit "safety" mix, it wouldn't hurt to apply dither. You'd be using it again to master to whatever end format you want. Any distortion you might get will accumulate faster than the noise added from dither. Dithering when you render the mix is your only chance to remove the distortion. As a final step when mastering, you can dither and apply noise shaping to move the noise away from the midrange where it's easiest to notice. You should only apply noise shaping once, so that would be when the target files will not undergo any further processing.

This seems like deja vu, pretty sure this has been asked/answered before... but don't most decent sound processing apps (like Sonar, Soundforge, Cubase, Logic Pro, etc) automatically dither when you downsample? How would you know for sure whether or not this is being done? What would be the effect of dithering something already dithered?
 
I'm not an expert on the subject, but I'm inclined to follow the advice of guys like James Johnston and Bob Ohlsson.

To paraphrase JJ, if you dither and you can hear an improvement, do it. If you dither and you're not sure if anything happened, do it. If you dither and it compromises the audio, something else in the system is broken.

One of Bob's pet peeves is guys who manufacture plugins that don't automatically dither their output. He advocates that dither is required at every step in processing and ideally we shouldn't have to think about it because it should be automatic. Unfortunately it does chew up significant processing power, so a lot of systems will have it as an option. Modern multicore processors have made this much less of an issue. Some of the best sounding digital systems available will allocate around 1/3rd of their processing to dither. "Triangular" or "TPDF" dither seems to be preferred.

Whether or not your system automatically dithers would hopefully be covered in its documentation. Ideally, it's automatic. Doesn't mean it is.

Especially in a 24 bit target format the source will play a role. Super clean acoustic music will make any low level distortion artifacts more apparent. Going to a 16 bit enviornment will make it much more apparent. Not sure if this is the case with Skrillex. Things like reverb tails, decaying notes and the spatial 3D perception of depth and width is what gets affected. Distortion from not dithering will stack around 3 times faster and is correlated to the signal. Working at a higher resolution to the target format like you're doing is a huge help on its own.

What we can percieve from it is a hugely subjective debate but if it doesn't hurt anything, there's no reason why we shouldn't.
 
I'm applying all of my master plugs on the stereo buss (mastering in the box) I've recorded everything in 24bit. I apply the cubase dither plugin (UV22) after my limiter and have the settings set to 16bit Hi. 'Hi' as in high resolution I assume....

I then export my track as a 16bit WAV.....

is this right?
 
I'm applying all of my master plugs on the stereo buss (mastering in the box) I've recorded everything in 24bit. I apply the cubase dither plugin (UV22) after my limiter and have the settings set to 16bit Hi. 'Hi' as in high resolution I assume....

I then export my track as a 16bit WAV.....

is this right?

Yes, if you want to ignore all the advice about mixing down to 24 bit with no processing on the master bus for archival reasons...
 
I'm not an expert on the subject, but I'm inclined to follow the advice of guys like James Johnston and Bob Ohlsson.

To paraphrase JJ, if you dither and you can hear an improvement, do it. If you dither and you're not sure if anything happened, do it. If you dither and it compromises the audio, something else in the system is broken.

One of Bob's pet peeves is guys who manufacture plugins that don't automatically dither their output. He advocates that dither is required at every step in processing and ideally we shouldn't have to think about it because it should be automatic. Unfortunately it does chew up significant processing power, so a lot of systems will have it as an option. Modern multicore processors have made this much less of an issue. Some of the best sounding digital systems available will allocate around 1/3rd of their processing to dither. "Triangular" or "TPDF" dither seems to be preferred.

Whether or not your system automatically dithers would hopefully be covered in its documentation. Ideally, it's automatic. Doesn't mean it is.

Especially in a 24 bit target format the source will play a role. Super clean acoustic music will make any low level distortion artifacts more apparent. Going to a 16 bit enviornment will make it much more apparent. Not sure if this is the case with Skrillex. Things like reverb tails, decaying notes and the spatial 3D perception of depth and width is what gets affected. Distortion from not dithering will stack around 3 times faster and is correlated to the signal. Working at a higher resolution to the target format like you're doing is a huge help on its own.

What we can percieve from it is a hugely subjective debate but if it doesn't hurt anything, there's no reason why we shouldn't.

Is that a yes or a no? That is, I get what you're saying but how does it help the OP?
 
Is that a yes or a no? That is, I get what you're saying but how does it help the OP?

I'm not sure. I was responding to Pinky.

It's basically an idea from a mastering engineer with around 5 decades of experience and a designer of bitrate reduced audio delivery formats about when and how to apply dither. I haven't read the Cubase manual, but I suspect the answer to the OP's question is there.
 
The Cubase manual does not answer the question.

Fairview, Boulder and Pinky gave the answers that the OP should consider.

Done.

My advice: Export 24-bit for mastering or future use of tracks-no dither. Dither to 16 bit if you don't care to go further and wish to share via MP3 or whatever streaming sites only.
 
I think Chili gave some good advice too. The one point that I would make is that you can't dither after the fact. Has to be when you render the mixdown.
 
Back
Top