Hello. I'm particularly interested in improving audio compression
using open standards (being a classical record producer, sometime
radio and tv producer and practicing musician) though I believe that
in a few years, we'll have better data storage and transfer
facilities that will render much compression un-necessary. Still,
for the time being, let's make it sound as good as possible!

Using the latest snapshot of pgcc for Linux (071598), I found that
compiling with -ffast-math caused an illegal MPEG3 bitstream to be
output, and compiling with anything more than -O2 (and
-march=pentium, -mcpu=pentium) caused the program to segfault (I'll
gdb it later to tell you exactly where).

I found that encoder worked in mono (altering the symbol in main.c
from MODE_STEREO to MODE_MONO) encoding a mono .wav file, and made a
good job of the sound down to 40kbs. By the way, the sound was much
better encoding a 32kHz sampled .wav at these rates than encoding a
48kHz .wav (same material, converted by the 'sox' program) which is
my audio Swiss army knife.

If I can help implementing/testing differeny psycho-acoustic models,
and particularly the MPEG-II layer III stuff, I'd be only too happy
to help. My experience in programming is limited (though I can
understand the code OK), debugging slightly better, using the GNU
compilers under Linux is fine, and I can do lots of listening! This
is the studio where I work.

John Hayward-Warburton
[EMAIL PROTECTED]

Reply via email to