Closed GoogleCodeExporter closed 9 years ago
Original comment by joe.wreschnig@gmail.com
on 5 Nov 2009 at 3:51
The following revised patch contains a test which will run provided sox, lame,
and
mp3gain are present. It also contains the original patch with a minor
modification.
Original comment by nondicti...@gmail.com
on 6 Nov 2009 at 3:58
Another patch, final one.
The test has been merged into test_mp3.py.
Track peak calculation code modified yet again to address portability concerns.
Original comment by nondicti...@gmail.com
on 6 Nov 2009 at 3:24
This seems like a needlessly complex test. We aren't testing ReplayGain from
first
principles - we're just testing whether Mutagen can read the tags LAME writes. A
couple of test files with known values would be a lot less fragile.
Original comment by joe.wreschnig@gmail.com
on 7 Nov 2009 at 2:58
Another patch. This time the test is much simpler, however, during testing I
discovered an apparent bug in LAME where encoding a 64kbps mono file with
22050Hz
sample rate would create a header frame that was one byte too small. This
resulted in
me having to rewrite some additional code.
The mp3 files are for the data directory.
Original comment by nondicti...@gmail.com
on 9 Nov 2009 at 11:42
Attachments:
Another revision. This one makes it possible to differentiate whether
track_gain is
meant to be 0 dB or if that data is not available, in which case the value will
be None.
Original comment by nondicti...@gmail.com
on 14 Nov 2009 at 12:41
Attachments:
Original comment by joe.wreschnig@gmail.com
on 25 Nov 2009 at 12:02
Reading this patch now, it's extremely branchy. I think to get this logic in,
the RG
parsing (and maybe the LAME/Xing/VBRI parsing in general) will need to be split
into
a separate function, and not have so many possible execution paths.
Original comment by joe.wreschnig@gmail.com
on 13 Dec 2009 at 4:14
[deleted comment]
[deleted comment]
[deleted comment]
Rather than split apart methods of MPEGInfo, I have instead rewritten it
entirely
from scratch.
The patch is against rev 73 and despite the rewrite incorporates the fix in
issue 46.
A couple of unit tests had to be altered due to the way the patched version
reports
the lengths of cropped files with VBR headers, and also the precise way in
which VBR
headers are detected by the patched version.
Original comment by nondicti...@gmail.com
on 8 Jan 2010 at 7:01
Attachments:
I'm very nervous about a rewrite of this code, especially one that removes all
the
docstrings and reference links.
I do not see the benefit of using a dict subclass to store frame information,
since
frame attributes are not at all freeform.
Original comment by joe.wreschnig@gmail.com
on 8 Jan 2010 at 7:46
A dict subclass was used for MPEGFrame so that MPEGInfo could pull out the
values to
be made public with a one liner.
Original comment by nondicti...@gmail.com
on 8 Jan 2010 at 8:29
Comments taken on board. Not sure how I can reassure though.
In addition this patch also reveals the VBR status which would close another
issue if
adopted.
Original comment by nondicti...@gmail.com
on 22 Jan 2010 at 9:32
Attachments:
Hi everyone, here's another take on the same subject, namely to give
(read-only) access to LAME's ReplayGain and peak tags in MP3s. This patch does
not do anything else, I tried to keep it as simple as possible. I also added a
simple test case and an accompanying test file.
Original comment by Vitos.La...@gmail.com
on 23 Jan 2013 at 7:24
Attachments:
mutagen has moved to Bitbucket: https://bitbucket.org/lazka/mutagen/issue/36
Original comment by reiter.christoph@gmail.com
on 4 Jul 2014 at 3:41
Original issue reported on code.google.com by
nondicti...@gmail.com
on 5 Nov 2009 at 1:21