32 bit rendering

[amigo]

Is it true you can render a 32bit file as many times you want without loosing any audio quality because of the rendering?

[bob katz]

Originally Posted by amigo

Is it true you can render a 32bit file as many times you want without loosing any audio quality because of the rendering?

That would be 32 bit float. If you are asking, "can I EQ first, save to a 32 bit float file, then compress, save to a 32 bit float file, then do another process without losing any audio quality?"

The answer is: The advantage of saving in a 32 bit float file is that you won't lose any more quality than you would if you had the plugins chained live. In other words, the output of the first plugin is 32 bit float, and so by saving it in that format, you don't lose anything and can then send that file later on to the next plugin you want to try.

However, don't disillusion yourself that there is no cost at all. Every DSP process creates some distortion and that distortion will accumulate. However, it will accumuate a lot more slowly by communicating in 32 bit float than by truncating to 24.

I'll also bet that if you optimized your level to be "near but a few dB below 0 dBFS", dithered to a 24 bit fixed point file and then sent that on to the next plugin, then audibly speaking it would be EXTREMELY hard to tell the difference compared to the 32 bit solution above.

To a great extent we are now beginning to argue "how many angels can fit on the head of a pin".

BK