This implementation uses decimation to generate an estimate of the
required ReplayGain adjustment for tracks sampled at high rates.
This approach avoids having to generate filters with commensurately more taps,
and also the subsequent effect on performance as these additional
taps are evaluated for high sample rate tracks.
Filter table entries with coefficients that are unchanged are
marked /* ORIGINAL */.
The remaining entries are new and have coefficient values obtained
from src/utils/loudness/loudness.sci. See:
http://lists.xiph.org/pipermail/flac-dev/2012-February/003220.html
Because these filter coefficients can be generated from a known source,
they are preferred to the FooBar2000 coefficients whose provenance is
unknown.
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
Fixes "format not a string literal and no format arguments
[-Werror=format-security]" errors.
Patch from Fabian Greffrath <fabian+debian@greffrath.com> via Debian.
Closes Debian bug #643377.
Fixes "undefined macro: AM_PATH_XMMS" and similar errors which cause
autoreconf to fail.
From Sebastian Andrzej Siewior <bigeasy@linutronix.de> via Debian.
Fix libFLAC.m4 may set empty -L flag when configure is called without
arguments, provided of course that the target package's configure script
is properly re-generated against the fixed libFLAC.m4
Patch from Fabian Greffrath <fabian+debian@greffrath.com> via Debian.
This change assumes that a C99 <stdint.h> header is available. For
compilers where that is not the case, the user should provide a
minimal replacement header.
Dave's comments:
This commit will break OS/2's EMX 0.9d library (GCC 2.8.1) which has been
been replaced by klibc. Considering the age of EMX and lack of testing
and that klibc contains so many improvements I think this is exceptable.
This disables the tests that don't run correctly out-of-tree from
the 'make check' target. Also add a new 'make fullcheck' target
which is the full old test suite.
This bug was causing the test_metaflac.sh test to fail after Earl Chew's
patch 19050f74eaa1aa6d609ca065e1a854ada5bb6b4c was supposed to fix an
older bug.
When reading the INDEX from the cue sheet, the format MM:SS:FF format
is disallowed if the sample frequency is not a multiple of 75 because
the index would only be approximate. However, 00:00:00 is _exact_
because it denotes the start of the track, so allow it as a special
case.
