ajh at atri.curtin.edu.au
Tue Oct 30 14:39:53 CET 2001
> did you use the method described on dsp-dimension?
No I used http://www.dspguru.com/ and some previous knowledge on
polyphase filter design.
> yes, i've also seen a strange thing, possibel cause of this.
> you only increment source data pointer by 1, but for >2x downsampling
> it requires increment by 2 or more.
> ok. i think it can be easily converted to MMX, so speed won't be problem.
> A'rpi / Astral & ESP-team
I have managed to get the down-sampling to run - it is not funny at
all. It will have to reside in a separate function (can not be mixed
with the up-sampling since the circular queue management is completely
different and it also needs pre-filtering of the incoming data). I
will create an example or add it directly to mplayer once the current
version is added to mplayer. That way my code will be consistent with
mplayers buffer management - and no one will need to re write it. Is
that ok with you?
In the meantime I will focus on optimising the filter parameters. I
think I have found a way to minimise the quantisation error by using
a type of genetic algorithm to tweak the filter parameters (really fun
By the way would it be interesting to add other sound features to
mplayer? I could do stuff like synthetic 4-6 channel stereo (synthesise
the missing channels from available data) or perhaps effects (room
expansion, pitch shifting (for karaoke), church effect etc)? Just an
More information about the MPlayer-dev-eng