Joined: 15 Mar 2007
|Posted: Fri Dec 14, 2007 1:12 am Post subject: volume between tracks
|I am putting together the material Gerard Bik had saved and another old
e-mail that I dredged up to describe exactly how I do the normalization
for a CD.
I got a small program "audio leak", which basically will very quickly
calculate the peak value of a track and the "A weighted average" as well
as the "unweighted average". As I understand it, the theory is that one
makes the A-avg the same for all tracks. I tried this for a CD that
contained one spoken voice track, some piano tracks -- without a great
deal of dynamic range -- and some vocal-piano tracks, with a lot of
The resulting CD using that program to decide how to set things (set of
course in Amadeus) was definitely better than using the peak values and
for the piano and voice-piano tracks it would have been acceptable. I
ended up taking the the spoken voice track down about 4.5 dB. I took
most of the piano tracks down about 2 dB.
Of course, some of my sentiments may be because I am primarily a vocal
person. I would say that the Audio Leak was definitely worth the money
(and the time) because it got me in the right ballpark very quickly.
added -- but you could probably use the rms average value from Amadeus
equally as well.
I ranAudio Leak on each file and put the peak and a-weighed value in a
spread sheet. I then figured out how much to amplify each one (using
dB) so that all had a constant dB-A value and the largest peak value was
I burned a CD, listened to it on a decent sound system and then listened
to adjacent tracks where I wasn't happy on Amadeus and moved the
relevant tracks down.
Burning the CD might not be necessary, but it let me listen to things in
a normal way, and also let me decide how I wanted to adjust the
To make everyting more specific, here are the numbers from several
tracks from that recording -- column 1 is the maximum value and track 2
is the A-weighted average; you could use the rms average for a track
with probably equally good results.
I took the difference between the peak sample value and the Leq value
The maximum difference is 23.2 (last one) So if the peak output of the
last one is zero dB, then the Leq value would be 23.2. I like to be a
bit conservative so I set the maximum value to -0.1 db and thus the Leq
value to be -23.3. So for the first track, I will need to "amplify" by
-1 dB, track 2 by 2.2. dB, etc. So I went back to Amadeus and amplified
each track by the appropriate dB value. This satisfies the criterion
that 1) everything ison scale and 2) you have the same Leq value.
So the amplfication values are
I hope that is all clear.
I suspect you long-time Amadeus list people are bored with the subject,
but I suspect you will see this same e-mail in 3-4 months when the
subject comes up again.
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list