#7 sfplay acts different on OSX/Intel and linux/PPC
Hi all, here's an interesting bug report.
I'm comparing the results of
cd regressionTests/audio
sfplay coll.mf -f out.au
(coll contains three files)
on OSX/intel and linux/PPC. (this is regression test #3) The two resulting out.au files are the same length, but the second and third sound files appear in different places within the file. It appears that the ending of each sound
file is zero-padded (to fill up the buffer size?), but the latter two sound files have samples switched around?
It's really weird; on linux/PPC, the first peak of the second sound file appears at sample 4565, whereas on OSX/Intel, the first peak of the second sound file is at sample 4578.
I'm not personally troubled by this bug (other than it screwing up the regression tests); the two files sound the same. But it might be an indication of more fundamental portability problems of marsyas.
Discussion
Anonymous
Logged In: YES
user_id=1618693
Originator: YES
stupid sourceforge; you should be able to upload two files at once. :(
File Added: osxintel.au
Logged In: YES
user_id=1618693
Originator: YES
Linux-i386 produces the same result as linux-ppc, so CPU architecture isn't an issue.
File Added: linuxintel.au