Posted by: "Lou Judson"
> As I understand it, if your peak level reaches -6dBFS you are using all
> the bits available. Lower than that and you are losing definition.
This has been the basic argument, but it may be flawed somewhat. People
tend to think a 1 in binary has significance and a zero does not. And
that's the source of this idea. The zeros are just as significant as the
ones. My son, who is a programmer laughs at folks on this.
The biggest problem is that sound levels are exponential and the coding
is not. In that way you do loose definition if you don't record fairly
close to the top of the scale. (just don't record at the top of the
scale or you will clip)
> the reference levels of -12 or -18 or -20 used in some conventions
> refers to Zero VU, not the maximum level...
When I refer to 0dB, I'm actually referring to that reading on the
portadisc's meters. I do know from actual experiment that reading is
really close to the clipping level if not on it. To be avoided always.
Because most metering is set up with a certain ballistics (to imitate
the behavior of older mechanical metering) there are peaks that exceed
what we see on our meters as they are too fast or too infrequent.
Avoiding clipping those is part of why we set to give some headroom on
the metering. On the portadisc, even the peak hold reading is not the
top of the sharpest peaks. I also do set based on expected callers and
caller distance. Cricket frogs, for instance may exceed the meter
reading by 10-15dB.
Walt
|