Hi there,

I've got a rather stupid question concerning the encoding process: Is there any sense in doing a 2pass encode with mppenc to feed the psy-model with optimum dc? I know that I can raise the gain later using replaigain - but if the gain level is too low, some part of the music that could be audible, will fall under the ath, right? So does it make sense to encode at first with, say --scale 20 so that mppend produces a warning message, and then encode again with the given gane value?

Northpack

Why do this? Can you hear any problems as it is? "Don't fix it if it ain't broken"...

/Erik

Hi ErikS,

I don't try to fix anything - I'm just aming for an optimal input signal. For example, I own some CDs which are old analoge recordings unkindly converted to a digital signal of approximately -15db about the whole album (max. gain, not headroom!). If I encode one of these albums with mpc without normalization, it'll be like raising the ath by 15 db. Or am I wrong? :confused:

Northpack

Don't these old recordings have a rather high noise floor? I mean: soft sounds that will be under the higher ath, would have been masked by this (noise floor) anyway.. no?

But I see your point. And I don't know the answer.