zdoom/fluidsynth
0
0
Fork 0
mirror of https://github.com/ZDoom/fluidsynth.git synced 2024-11-27 22:43:36 +00:00
Code Issues Projects Releases Packages Wiki Activity

Fix SF3 soundfont loading (uses byte positions for sample start/end, not sample word positions)

Browse source
This commit is contained in:
Marcus Weseloh 2018-03-24 18:31:33 +01:00
parent 945214b582
commit f5f85a1dfe
1 changed files with 12 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
src/sfloader/fluid_defsfont.c
View file

@ -3367,11 +3367,16 @@ fixup_sample (SFData * sf)
int invalid_loops=FALSE;
int invalid_loopstart;
int invalid_loopend, loopend_end_mismatch;
unsigned int total_samples = sf->samplesize / FLUID_MEMBER_SIZE(fluid_defsfont_t, sampledata[0]);
unsigned int total_bytes = sf->samplesize;
unsigned int total_samples = total_bytes / FLUID_MEMBER_SIZE(fluid_defsfont_t, sampledata[0]);
p = sf->sample;
while (p)
{
/* Standard SoundFont files (SF2) use sample word indices for sample start and end pointers,
* but SF3 files with Ogg Vorbis compression use byte indices for start and end. */
unsigned int max_end = (sam->sampletype & FLUID_SAMPLETYPE_OGG_VORBIS) ? total_bytes : total_samples;
sam = (SFSample *) (p->data);
/* ROM samples are unusable for us by definition, so simply ignore them. */
@ -3384,8 +3389,10 @@ fixup_sample (SFData * sf)
/* If end is over the sample data chunk or sam start is greater than 4
* less than the end (at least 4 samples).
* FIXME: where does this number 4 come from? */
if ((sam->end > total_samples) || (sam->start > (sam->end - 4)))
*
* FIXME: where does this number 4 come from? And do we need a different number for SF3 files?
* Maybe we should check for the minimum Ogg Vorbis headers size? */
if ((sam->end > max_end) || (sam->start > (sam->end - 4)))
{
FLUID_LOG (FLUID_WARN, _("Sample '%s' start/end file positions are invalid,"
" disabling and will not be saved"), sam->name);
@ -3400,8 +3407,8 @@ fixup_sample (SFData * sf)
* this is as it should be. however we cannot be sure whether any of sam.loopend or sam.end
* is correct. hours of thinking through this have concluded, that it would be best practice
* to mangle with loops as little as necessary by only making sure loopend is within
* total_samples. incorrect soundfont shall preferably fail loudly. */
invalid_loopend = (sam->loopend > total_samples) || (sam->loopstart >= sam->loopend);
* max_end. incorrect soundfont shall preferably fail loudly. */
invalid_loopend = (sam->loopend > max_end) || (sam->loopstart >= sam->loopend);
loopend_end_mismatch = (sam->loopend > sam->end);