Phil Leigh wrote:
There is a problem with ANY kind of digital normalisation, in that
unless it is a pretty dumb "auto-gain control" it may introduce
compression - otherwise a quiet track might be boosted so much that the
loudest parts clip.
I agree that normalization is generally dumb, but it never introduces
compression.
Normalization means taking the existing signal and making it be as loud
as possible in a 16 bit PCM signal. in 16 bits, you can hold 0 to 64K,
but in PCM, it is +32K to -32K
So if you have a signal that has a max of only 4K, to normalize it, you
multiply all signals by 8. This raises the max signal to 32K, and raises
any noise signal by 8 as well. It does nothing to the signal to noise
ration.
Compression takes a loud signal and makes it smaller.
It does _not_ take a quiet signal and make it louder. But
most people use compression with make up gain. Thus when you take the
loud signals and make them smaller without changing the small signals,
and then make all signals as loud as the loudest used to be, you get the
impression that loud sounds stayed the same, and quiet parts are louder
So, if it doesn't boost or cut the track by a simple
fixed amount of dB, it is compressing and that is a BAD thing since it
is sucking the original dynamics out.
No, see above
...and if it is a dumb auto-gain shifter...how does it avoid clipping?
By scanning the file, finding the highest value, and adjusting to suit.
I think replaygain does have a use for casual listening, parties
etc...but I'd never use it when I really want to hear what the artist
intended.
Or more precisely, what the producer and recording engineer and
mastering engineer wanted you to think that the artist intended.
--
Pat Farrell PRC recording studio
http://www.pfarrell.com/PRC
_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/audiophiles
