Re: RFS: splay -- Sound player for MPEG-1,2 layer 1,2,3

[Goswin von Brederlow <brederlo@informatik.uni-tuebingen.de>]

John Hedges <john@drystone.co.uk> writes:

> On Tue, Jul 06, 2004 at 05:51:25PM +0200, Andreas Metzler wrote:
>> On 2004-07-06 John Hedges <john@drystone.co.uk> wrote:
>> > On Tue, Jul 06, 2004 at 02:45:17PM +0200, Andreas Metzler wrote:
>> > > On 2004-07-06 John Hedges <john@drystone.co.uk> wrote:
>> > > > I'd like to adopt splay, a command line mp3 player. I've updated the bug
>> > > > report (#246971) to reflect ITA but have a couple of questions before
>> > > > uploading.
>> > > [...]
>>  
>> > > I have not used and tested it but have just looked at package
>> > > description and website. How does splay differ from mpg321?
>> 
>> > splay is the fastest mp3 decoder I know of, using about one third the cpu
>> > required by mpg123 (based on woody packages).
>> 
>> I see.
>> 
>> You are mixing up mpg123 and mpg321 BTW. The former is fast and
>> non-free, the latter is slow (libmad, no FPU) and free. To make it
>> easy to mix this up /usr/bin/mpg123 is managed by dpkg-alternative,
>> i.e. invoking mpg123 will start mpg321 if installed.
>> 
>> > I just tried mplayer so now know of a faster one :)
>> 
>> mplayer's MP3 decoder mp3lib is based on mpglib which is a part of
>> mpg123. While mpg123 is nonfree, its mp3-decoding engine is not.
>> (xmms also uses it.) Sadly I do not know a free commandline mp3-player
>> in Debian based on mpglib.
>> 
>> > This makes splay more suitable than mpg123 when decoding to files or
>> > when using slower hardware. Mainly I use epia based MiniITX as mp3
>> > players, but the tests below are on a P4:
>> 
>> >     john@desk:~$ time mpg123 -s 1.mp3 > /dev/null 2>&1
>> 
>> >     real    0m6.278s
>> >     user    0m6.260s
>> >     sys     0m0.000s
>> 
>> I bet this slow "mpg123" is a mpg321. ;-)
>>                   cu andreas
>
> Quite right - I did use mpg321 but wasn't aware of the distinction. I
> remember reading of licensing problems resulting in mpg321 but had no
> idea it was based on an entirely different decoder.
>
> However it might be that the splay decoder is comparable to that of
> mp3lib. My quick and not so reliable tests showed it to be about 15%
> slower but that was comparing a stock i386 splay with an optimised i686
> mplayer.
>
> I'm more than happy with splay and have used it for years. Well worth a
> sponsor - I'm sure you'll agree ;)
>
> Cheers
>
> John

Maybe you can compare them a bit more and if mpglib is faster or
comparatively the same you could port splay over to that
library. Better to have common code for a common problem.

MfG
        Goswin