I am doing Cassette to CD transfers and this question emrges from
that: What is the rule of thumb (if there is one) for the volume to
aim for normalization of a song collection on CD's? Is it -18 dB or
is it -16 dB? It seems, reading some web sites and Goodle archives,
that one just normalizes all songs on a CD collection to the same
level; what that level is not that important. But is this really
sufficient? Isn't there a standard level to make them same as what is
mostly accepted as OK volume. (I mean if you give your CD to a
friends, they would neither feel it is too loud, nor too quiet.)

Thanks.

-surinder

PS: Now that we are on the subject of loudness, what is the thing that
measures loudness, "average RMS" or "total RMS"? (I see this in Sound
Forge). I know what RMS is, but I cannot fathom what average or total
refer to; RMS itself is am averging.

SPS22 wrote:
I am doing Cassette to CD transfers and this question emrges from
that: What is the rule of thumb (if there is one) for the volume to
aim for normalization of a song collection on CD's? Is it -18 dB or
is it -16 dB? It seems, reading some web sites and Goodle archives,
that one just normalizes all songs on a CD collection to the same
level; what that level is not that important. But is this really
sufficient? Isn't there a standard level to make them same as what is
mostly accepted as OK volume. (I mean if you give your CD to a
friends, they would neither feel it is too loud, nor too quiet.)

If you are transferring a cohesive cassette, find the loudest peak on tall
the tracks, normalise that (peak, not rms), then raise all the other tracks
by the same amount. That way you will retain the mastering decisions re
relative track volumes.

If random tracks, then it's up to you. Average RMS of -12dB is loud. But
using 'average RMS' can be a bit hit and miss regarding producing overs,
unless you use an RMS normalisation app with compression, and then it's
totally out of your control.

As cassette material is likely of low dynamic range anyway, just peak
normalise each track, reducing any that stick out too loud.


geoff

When you normalize, you are really just changing the amplitude. You
usually adjust it to maximaze the amplitude you can without going into
saturation. You should normalize to 100% (0 db), or maybe 97-98%
,because maybe the program you use to normalize might let a bit of the
signal go a touch over. The quietness or loudness of a song is relative
to many factors. Normalizing the CD would eliminate one of them, and
that would be a good thing. The other factors include how high the
volume is set on the amplifier, how loud the song is compared to other
songs on the CD, etc. If you nomralize each song separately on the same
CD, then that's the best thing.

CD


SPS22 wrote:

I am doing Cassette to CD transfers and this question emrges from
that: What is the rule of thumb (if there is one) for the volume to
aim for normalization of a song collection on CD's? Is it -18 dB or
is it -16 dB? It seems, reading some web sites and Goodle archives,
that one just normalizes all songs on a CD collection to the same
level; what that level is not that important. But is this really
sufficient? Isn't there a standard level to make them same as what is
mostly accepted as OK volume. (I mean if you give your CD to a
friends, they would neither feel it is too loud, nor too quiet.)

Thanks.

-surinder

PS: Now that we are on the subject of loudness, what is the thing that
measures loudness, "average RMS" or "total RMS"? (I see this in Sound
Forge). I know what RMS is, but I cannot fathom what average or total
refer to; RMS itself is am averging.

On Thu, 29 Jan 2004 16:25:05 +1300, "Geoff Wood"
-nospam wrote:

SPS22 wrote:
I am doing Cassette to CD transfers and this question emrges from
that: What is the rule of thumb (if there is one) for the volume to
aim for normalization of a song collection on CD's? Is it -18 dB or
is it -16 dB? It seems, reading some web sites and Goodle archives,
that one just normalizes all songs on a CD collection to the same
level; what that level is not that important. But is this really
sufficient? Isn't there a standard level to make them same as what is
mostly accepted as OK volume. (I mean if you give your CD to a
friends, they would neither feel it is too loud, nor too quiet.)

If you want to make your CD collection have a similar loudness to
commercially-released CDs, then you will have to both compress the
dynamic range and peak normalise to 0dB (or maybe -0.1dB to avoid
possible playback problems with sub-standard DACs). The normalisation
bit is easy. The choice of compression factors is not; you'll have to
do some experimentation.

If random tracks, then it's up to you. Average RMS of -12dB is loud.

I was about to say that a peak level of -12dB is very quiet, but if
the OP was talking about RMS levels, then yes, -12dB is fairly loud,
but by no means as loud as some recent commercial releases.

As cassette material is likely of low dynamic range anyway, just peak
normalise each track, reducing any that stick out too loud.

Actually, typical vintage cassette material is likely to have a
*greater* dynamic range than modern pop/rock CDs. This is a very sad
reflection of how modern mastering methods have destroyed any
semblence of natural dynamics, all for the sake of "everything louder
than everything else".
--
Clive Backham

Note: As a spam avoidance measure, the email address in the header
is just a free one and doesn't get checked very often. If you want to email
me, my real address can be found at: www [dot] delback [dot] co [dot] uk

"SPS22" wrote in message
om

I am doing Cassette to CD transfers and this question emerges from
that: What is the rule of thumb (if there is one) for the volume to
aim for normalization of a song collection on CD's?

Wrong question. Normalization is not a good idea if quality results are your
goal.

Is it -18 dB or is it -16 dB?

Wrong question. Numbers that small would relate to average levels, which
ignores the fact that in a digital system, you first and foremost don't want
any clipping.

It seems, reading some web sites and Google archives,
that one just normalizes all songs on a CD collection to the same
level; what that level is not that important.

And when you play back the songs, some are subjectively too loud and some
are subjectively too soft. This is because the natural peak-to-average ratio
of music can vary by quite a bit.

But is this really sufficient?

Depends what your goal is. If you want your songs to have a similar
subjective loudness when you play them back, then you would be wise to use
your ears, or find something that tries to perform like an ear.

Here are some examples of programs that try to respond to loudness like the
human ear:

http://www.measure.demon.co.uk/Acous.../loudness.html

http://replaygain.hydrogenaudio.org//software.html

http://www.xs4all.nl/~walterln/replaygain/

Isn't there a standard level to make them same as what is
mostly accepted as OK volume. (I mean if you give your CD to a
friends, they would neither feel it is too loud, nor too quiet.)

PS: Now that we are on the subject of loudness, what is the thing that
measures loudness, "average RMS" or "total RMS"? (I see this in Sound
Forge). I know what RMS is, but I cannot fathom what average or total
refer to; RMS itself is am averaging.

Right, but when you are measuring something that varies like music, you do
the measurement over a period of time. This period of time can be
user-selected or it can be the whole song. Within that period of time,
programs can break the period down into shorter little pieces and measure
each little piece.

Average RMS statistics relate to the entire time period that was measured.
Peak, and minimum RMS statistics relate to measurements based on the shorter
little pieces. One of those little pieces is going to be the one with the
highest statistics, and one of them is going to be the one with the lowest
statistics.