Hi, Joachim.
On Sep 16 2007, Joachim Reichel wrote:
> do you use the normalize-mp3 script or do you perform
> decompression/normalization/compression on your own?
No, I'm not interested in the script. I use decompression, normalization
and compression, mostly because some of the files that I have are not
even in MP3 format. Some are in other formats (some lossless).
I have been using shntools's shnjoin to join the uncompressed wave files
and I was able to reproduce the problems even with an Intel machine
running MacOS X, which leads me to think that the problem may be
upstream.
I tried to read the source to see if I could easily spot the problem,
but it was taking too much time and I therefore have "paused" my reading
of the source code. :-(
> If you do not use the script: Do I understand you right, that the
> normalize command exits without error message, with error code 0, and
> the resulting wave file is shorter than the orginal?
I didn't see the return code of normalize, unfortunately, but it didn't
print any error message to stderr (which one would assume in case of
error, of course).
> Note that quite much disk space is needed in the process. Is it
> possible that you run out of disk space and is this is somehow not
> detected?
No, I have enough inodes for use (I have only 1% of them used) and
enough disk space for decompressing files (an external HD with 200GB
left for me to use).
> How large is the uncompressed wav file? Is it larger than 2GB or 4GB?
No, when the files are uncompressed, and joined together, they take
approximately 600MB, which is about the standard length of an audio CD.
> Please let me know the exact commands including command line options
> that you use.
For decompressing, say, mp3 files, I am using madplay. For joining the
resulting wave files, I am using shntools's shnjoin. Then, I normalize
the audio with a simple normalize call and split the files again with
shnsplit.
The exact commands for decompressing the files are listed here:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
#!/bin/sh
while read file; do
file=${file##/Volumes/Backup/mp3/}
directory=$(dirname "$file")
filename=$(basename "$file" .mp3)
mkdir -p "$directory" || true
cd "$directory"
madplay -a -3 -S -v -o "$filename.wav" "/Volumes/Backup/mp3/$file"
cd ../..
done
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
And for recompressing are here:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
#!/bin/sh
while read line; do
cd "$line"
shnjoin [0-9]*.wav
shncue [0-9]*.wav > output.cue
normalize-audio joined.wav
cuebreakpoints output.cue | shnsplit joined.wav
for i in split-track*.wav; do
echo $line
lame --noreplaygain -V 2 --vbr-new $i;
done
cd ../..
done
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Just after the call to normalize-audio I notice that I have problems
with the file and that it does not always contains the same number of
samples that it once contained. :-(
I have tested this doing this manually and compressing the file with
flac just before the normalize-audio step above and compressing it again
and flac complains that the file ended prematurely... :-( shnsplit also
gives me the same error message (with other words). :-(
Please, again, notice that this does not happen *every* time and, thus,
this is the reason why I have not posted this before. But I found some
albums that I have that *do* make normalize-audio exhibit the behavior
cited before. :-(
Thanks, Rogério Brito.
--
Rogério Brito : [EMAIL PROTECTED] : http://www.ime.usp.br/~rbrito
Homepage of the algorithms package : http://algorithms.berlios.de
Homepage on freshmeat: http://freshmeat.net/projects/algorithms/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
