On 15.10.2012 23:55, Torne Wuff wrote: > I could easily be wrong, I don't know how our dsp system works very > well :) I looked in dsp_replaygain_update in > lib/rbcodec/dsp/dsp-misc.c and it looks like it just takes the peak > value from the metadata and checks if the replaygain value would put > it over, and if so makes it smaller. This appears to get called on > track change and on replaygain settings change, then it just sets it > as the static gain value applied on decoding. If that is correct, the previous argument of songs of the same album become desynchronized in volume is invalid, as replaygain also features a Albumwide-Peak-Value, at least if created with metaflac.
cheers, -- Jonas