djwayne wrote on Sat, 07 May 2005 18:56 |
Correct me if I'm wrong, but it's my understanding that when you record with a 24 bit sound card, all you're getting is a 24 bit file, the additional 8 bits aren't used at all. They remain zero's. Applying effects to the 24 bit file, come out a little smoother sounding, than if they were applied to a 16 bit file. After the effects desired are applied, it all gets squashed down to 16 bit to go onto a cd, or left at 24 bit and recorded to dvd.
|
Wayne,
The moment the file enters the software on playback it becomes 32-bit float, 24-bits with 8 extra zeros. Those zeros stay zeros until you twist a knob or move a fader or do some other dsp operation. When that happens they become "active" or in other words they contain data that is used by the dsp. They aren't zeros anymore. Using (yes actually using) all 32-bits allows a lot better math and better dsp operation.
The reason for using 32-bit float internal is that more accurate mathematical calculations can be done which results in better quality. 32-bit floating point math can represent numerical figures that 24-bit math cannot. 24-bit might get close and round the math off. That's the reason why Steinberg and everyone else uses (at least) 32-bit float math internally. This includes the Mackie d8b and probably a lot of other digital hardware mixers that use 32-bit float for dsp.
After the numbers are crunched and the audio is output it is truncated or dithered back to 24. The dsp effected 24-bits that remain sound better than they would otherwise because they were processed with that "higher" 32-bit math. I think all daw and 2-track editing apps use at least 32-bit float for internal processing some use double precision (PTHD) and a couple of 2-track apps use 64-bit internal.
Just remember the bigger the internal bit number the better (more accurate) the math can be, the better (again more accurate) the overall dsp sound. No matter if it eventually ends up as a 24 or 16-bit file. This is while you're "twisting" the audio.
That's the theory anyway. DAW's do sound much better now than when they were processing at 24-bit internal. They all do at least 32-bit float internally so the idea that Steinberg did it "so they can record 32-bits" is not true. They did it to make their daws sound better.
Again (just to be clear) I can't take credit for "knowing" any of that. It was explained to me by Nika Aldrich. I'm not good with math.
Recording tracks at 32-bits? Dunno. And what you said is true, unless you render those tracks with dsp those 8 zeros in the files will remain 8 zeros in the files. Waste of disk space unless you're recording with effects during tracking.
The real benefit in Nuendo/SX is in using 32-bit float mix files as masters (rendered, not recorded through output converters). No truncation.
Lawrence