[LAD] peak has nothing to do with loudness (check out EBU R128)

Jörn Nettingsmeier nettings at folkwang-hochschule.de
Mon Dec 20 18:47:05 UTC 2010

On 12/20/2010 07:29 PM, Paul Davis wrote:
> 2010/12/20 Jörn Nettingsmeier <nettings at folkwang-hochschule.de>:
>>> there is a complication: loudness is no identical to maximum sample
>>> value, but the relationship is good enough for government work, so to
>>> speak.
>> no, it's not. there is absolutely no meaningful relation whatsoever
>> between perceived loudness and peak value.
> there is if they are all samples of the same instrument.
> that's why i said "good enough for government work" :)

ok, i took that as a general remark...

but still: chances are multiple dynamic variants of the same instrument
have the same peak value, if the maker used peak limiting to avoid s/n
issues on very dynamic instruments (which is quite likely).

and you will not be able to map multiple sounds to perceptually equal
loudness levels. say you want to create a "drum map" for your keyboard:
you will want the toms to be just as loud as snare and bass drum when
you play them with "average" velocity.

maybe the "replay gain" measurement features present in some audio
players could be used for the job (perhaps some are even scriptable)?
i'd expect them to have a very long time window though, so it's probably
best to feed them 10-30 repetitions of the same sound for the
measurement system to stabilize.

More information about the Linux-audio-dev mailing list