At 2007-05-13 08:24, Lou Judson wrote:
>There is much to "lose" by not using 24 bit - it is not signal to
>noise per se, but finer gradation of detail, and a wider range of
>processing that can be done with 24 bit.
>
>The point of the articles you quote is that if the preamp noise is so
>bad then 24 bit won't sound better, but that is wrong. 24 bit gives
>far more definition, especially if any processing will be done to the
>audio - even just a simple change in level or a fade out. And any
>software attempt to remove noise will be more successful in 24 bit.
>It is somewhat similar to the difference between cassette and reel
>tape in my experience.
But, the same depth of processing would be possible after a
16-to-24-bit conversion. As long as your noise floor is significantly
louder than -90 dB (20*log(1/2^(16-1))), there will be no significant
difference, except for the savings in storage space with 16 bit.
The noise floor is only going to be on the order of -90 dB if you
have the gain set low in order to get extreme dynamic range (catching
the subtleties of a very quiet background ambience while also
capturing very loud intermittant sounds) or to avoid ever needing to
change your recording volume (which could be useful for science, i.e.
measuring absolute loudness of sounds recorded).
|