Hi, If I’m tracking at 16 bit in my project do I need to add dithering when mastering, I believe that if I was tracking at 24 bit I would need to dither down to 16 bit before export but if I’m already at 16 bit is this necessary?
Dithering is simply introducing noise at low level, in order to conceal the high levels of distortion present in digital signals as they approach silence (typically at the end of the fade or reverb tail). If you listen to this at high levels (careful!) you will hear a 16bit audio recording break up into grainy distortion just before it disappears. After dithering, the signal will disappear into a soft noisy haze, which is considerably more pleasant, particularly if you like to listen to the ends of fades at very high volume!
So yes, I’d say your 16bit recording would benefit from dithering. Dithering isn’t important if you’re staying in the 24bit world, because the digital noise floor is so much lower.
If you had all your audio at 16 bit and were working at 16 bit, then you would not need to dither, as dithering is only used when reducing the quantization bit level,
but…
even though all your audio may be at 16-bit, Cubase is working internally at 32-bit, so yes, you should dither to 16-bit when exporting from Cubase, in the same way that if you work at 24-bit, you should dither down to 24-bit when exporting, because Cubase has been working at 32-bit.
Actually, that 32 bit is still 24bits of mantissa and 8bits of exponent.
However, if 0dB has been exceeded at any time up till output, those ‘overload’ samples will have lower bits that actually match the not so low bits of the other samples, due to the increase in the exponent shifting them all up, so dithering is still recommended.
I think you are confusing dithering with floating point. Floating point exports will retain the ‘overs’ in tact without clipping them, but at some point you will need to limit and/or reduce the level. Dither should not be used when exporting a 32 bit file, and will NOT make an audible difference even with 24 bit exports. Dither should be used when exporting a 16 bit file (even if your tracks are 16 bit as the mixer is floating point as mentioned earlier), but even then the improvement is small. The importance of dither is often overstated and loads of incorrect information about it are posted on internet forums.
No. What I was getting at is that the nth bit of a floating point mantissa bits may not represent the same absolute value from sample to sample, which means the absolute digital noise floor moves up and down.
However, I would agree that at 24 (or 23 or 22) bits, dithering is a non-issue. It is probably a non-issue at 16 bits for most music, except for sparse or very dynamic genres, where the digital noise floor might be perceived in the softer spaces in a quiet room.
PS. If you are using some sort of “correct” tape machine emulation plugin, you will not need to dither. The noise within the tape circuit is a “kinf of” self dithering (dithering is noise, somewhat controlled).
This should be the same as running your mix through an analogue tape in mastering.
24 bit out > through tape > 16 bit into Master recorder, is not unheard of in the professional mastering world (you don’t need to add noise to noise )
Please read upon this, or consult someone who knows this stuff, before nagging me on this. You know who you are