I'd try testing a build without optimisation to see if its a
	compiler opt problem.

On Tue, 1 Jan 2002, Stefan Wuerthner wrote:

> I succesfully built madplay (mad-0.14.2b) from the sources on a DNARD shark.
>
> After that I converted several mp3 files to wave, but unfortunately some of
> the resulting wave files sound horrible.
>
> There is certainly no problem with the input files, and an additional
> conersion test with the Windows executable of mad-0.14.2b revealed that the
> fault lies definitely within the NetBSD/arm32 binary of madplay.

-- 
		David/absolute		-- www.netbsd.org: No hype required --