32-bit, 24-bit, etc.

rpmbassman wrote on 3/25/2003, 1:39 PM
What is the internal processing bit depth in CDA? It says it supports 32-bit, but is everything computed using 32-bit floating-point?

For example: if I cross-fade two 16-bit wave files, is the cross-fade computed using 32-bit?

If I do nothing but append 16-bit wave files, do I need to dither because they are cnverted to 32-bit first? I would hope not. They have already been dithered to 16-bit.

If I have a mix of 32-bit, 24-bit and 16-bit wave files, should I just apply the dithering to the 32-bit and 24-bit tracks? I would hope so.

I see no indication of bit-depth in CDA. It would be nice if this were displyed somewhere.

Comments

Geoff_Wood wrote on 3/26/2003, 12:43 AM
16 bit files are not necessarily 'converted to 32 bit' when being used in a 32 bit processing environment. Think of it as 'headroom'.

geoff
rpmbassman wrote on 3/26/2003, 8:57 AM
I understand that. What I meant was: are they put in the 32-bit format, thereby necessitaing dithering. If I'm not doing any processing on the 16-bit media, I shouldn't have to dither it.
Geoff_Wood wrote on 3/27/2003, 2:00 AM
No dithering is required for 16 bit signals to exist in a 32 bit environment. They occupy the bottom 16 bits 'unchanged'. That is until any processing is done. Unless I'm wrong...


geoff
Geoff_Wood wrote on 3/27/2003, 10:33 PM
.... and of course, dithering is applied only to the processed area (crossfades in this instance).

geoff
jorgensen wrote on 3/28/2003, 2:54 AM
I have had the same though, but now do not concern.

CDA uses 32-bit floating point internal (I do not believe it uses 64-bit), and only if an (old) FX do not support floating point, it is truncated to 24-bit and back again.

Therefore I have done some test with and without dithering on the output signal, and I simply cannot hear a difference on normal materials. Only if I use special materials, lower the level considerable and use a special dithering FX, I can hear an slightly improvement.