21 Sep
2009
21 Sep
'09
11:17 p.m.
I have read that it is good practice to aim for a maximum level of -3.0 dB when mixing,
mastering, normalizing etc. in the digital domain.
Why is this so? And are there situations when one might aim for 0 dB?
TIA
Norv
__________________________________________________________________________________
Get more done like never before with Yahoo!7 Mail.
Learn more: http://au.overview.mail.yahoo.com/
21 Sep
21 Sep
11:20 p.m.
On Mon Sep 21, 2009 at 06:17:04PM -0700, Norval Watson wrote:
I have read that it is good practice to aim for a
maximum level of -3.0 dB when mixing, mastering, normalizing etc. in the digital domain.
two mono chans at -3 should sum to 0db, afaik
Why is this so? And are there situations when one
might aim for 0 dB?
TIA
Norv
__________________________________________________________________________________
Get more done like never before with Yahoo!7 Mail.
Learn more: http://au.overview.mail.yahoo.com/
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
http://lists.linuxaudio.org/mailman/listinfo/linux-audio-user
22 Sep
22 Sep
12:04 a.m.
> I have read that it is good practice to aim for a
maximum level of -3.0 dB
when mixing, mastering, normalizing etc. in the
digital domain.
two mono chans at -3 should sum to 0db, afaik
OK, and thanks for quick reply Carmen...
In Ardour specifically, I have a stereo master bus with L and R meters each hitting about
-3 dB.
The single numeric display above the meters also reads about -3 dB, rather than 0 dB.
Is this what I should be aiming for?
Norv
__________________________________________________________________________________
Get more done like never before with Yahoo!7 Mail.
Learn more: http://au.overview.mail.yahoo.com/
6:52 a.m.
Norval Watson wrote:
in theory, it's ok to go all the way to full scale (-0dB).
in practice, there are a few problems:
* cheap D/A converters can start to sound nasty even when not overloading.
* it's hard to check for overload at 0dB - you can't ever get louder
than full scale in fixed-point notation, so an overload can only be
deduced by a number of consecutive samples at full scale
* you can have a valid signal with two samples at 0dB, but that implies
that there is an inter-sample peak between those two that is larger than
that. depending on the rest of the signal chain, this can cause problems
(think over-sampling).
for pop productions, -3db is on the conservative side - most people seem
to go for less than -1. for classical, many tonmeisters will not
normalize at all, so that the natural absolute loudnoess is retained (to
some extent) - i.e. you set your volume so that you can just about
endure that wagner orchestra cd, and then you put that lute music on,
and it will peak at -12dB or so... so you don't have to readjust the
volume so much.
> I have read that it is good practice to aim
for a maximum level of -3.0 dB
when mixing, mastering, normalizing etc. in the
digital domain.
two mono chans at -3 should sum to 0db, afaik
OK, and thanks for quick reply Carmen...
In Ardour specifically, I have a stereo master bus with L and R meters each hitting about
-3 dB.
The single numeric display above the meters also reads about -3 dB, rather than 0 dB.
Is this what I should be aiming for? 
8:22 a.m.
On Mon, Sep 21, 2009 at 06:19:51PM -0700, carmen wrote:
On Mon Sep 21, 2009 at 06:17:04PM -0700, Norval Watson
wrote:
They could sum to anything between -inf and +3dB,
depending on how related they are.
And even for decorrelated signals, where RMS
of the sum will 3dB higher than each channel
separately, the peak level will increase by
more than 3dB, up to 6dB.
What is the 'correct' level for any recording depends
on the type of music, the context, where and how the
recording is going to be used, etc.
If you are participating in the loudness war (see
http://en.wikipedia.org/wiki/Loudness_war) you will
try to squueze out the last fraction of a dB.
Usually with disastrous effect on the quality of
the sound or music, even if it does not clip or
distort.
If you go for a more musically enjoyable result it
depends very much on the type of music. For anything
'heavy' you'd want to maximise the loudness without
clipping. In that case a fast peak limiter at -1dB
or so will do wonders without affecting the sound
too much.
For anything classical or when you want to preserve
the natural dynamics of real instruments and performers.
peak levels are less of an issue, and your decisions on
level should be based on what your ears tell you, and be
guided by a meter that does indicate both RMS and peak.
Ardour's meters are near to useless for this, jkmeter
will do a good job if you learn to use it.
Ciao,
--
FA
Io lo dico sempre: l'Italia è troppo stretta e lunga.
I have read that it is good practice to aim for a
maximum
level of -3.0 dB when mixing, mastering, normalizing etc.
in the digital domain.
two mono chans at -3 should sum to 0db, afaik 
9:08 p.m.
----- Original Message ----
From: Fons Adriaensen <fons(a)kokkinizita.net>
On Mon, Sep 21, 2009 at 06:19:51PM -0700, carmen
wrote:
Thanks Fons !
I believe I am still thinking in terms of 4-track cassette recording where a
"hot" signal is all-important to minimise the dreaded tape-hiss ...
The music I am mixing is radio, iPod-friendly trip-hop with 95% synthetic instruments, and
the vocals have a lot of effects as well.
Norv
__________________________________________________________________________________
Get more done like never before with Yahoo!7 Mail.
Learn more: http://au.overview.mail.yahoo.com/
On Mon Sep 21, 2009 at 06:17:04PM -0700, Norval
Watson wrote:
They could sum to anything between -inf and +3dB,
depending on how related they are.
And even for decorrelated signals, where RMS
of the sum will 3dB higher than each channel
separately, the peak level will increase by
more than 3dB, up to 6dB.
What is the 'correct' level for any recording depends
on the type of music, the context, where and how the
recording is going to be used, etc.
If you are participating in the loudness war (see
http://en.wikipedia.org/wiki/Loudness_war) you will
try to squueze out the last fraction of a dB.
Usually with disastrous effect on the quality of
the sound or music, even if it does not clip or
distort.
If you go for a more musically enjoyable result it
depends very much on the type of music. For anything
'heavy' you'd want to maximise the loudness without
clipping. In that case a fast peak limiter at -1dB
or so will do wonders without affecting the sound
too much.
For anything classical or when you want to preserve
the natural dynamics of real instruments and performers.
peak levels are less of an issue, and your decisions on
level should be based on what your ears tell you, and be
guided by a meter that does indicate both RMS and peak.
Ardour's meters are near to useless for this, jkmeter
will do a good job if you learn to use it.
Ciao,
--
FA I have read that it is good practice to aim for a
maximum
level of -3.0 dB when mixing, mastering, normalizing etc.
in the digital domain.
two mono chans at -3 should sum to 0db, afaik
5594
days inactive
5595
days old
linux-audio-user@lists.linuxaudio.org
6 comments
5 participants
participants (5)
-
carmen -
Fons Adriaensen -
Jörn Nettingsmeier -
Norval Watson -
Ray Rashif