[MP3 ENCODER] floating point accuracy

[Michael Cheng]

hi all,
        some random thoughts...

Tord (bladenc author) claims that FP accuracy is part of the reason some
ISO implementations don't sound that good.
Who was it that changed all occuracnes of 'float' to "FLOAT" and then did
"#define FLOAT double"?  Peman?  Did you get a speedup? slowdown? quality
change?  Anyone have any opinions?

later
mike