Subject: RE: [linux-audio-dev] MuCoS, Glame, API Issues
From: Karl W. MacMillan (karlmac_AT_peabody.jhu.edu)
Date: ti maalis 07 2000 - 21:41:27 EST
>
> It should be no problem to write a wrapper for ladspa plugins. But I just
> like to know how you handle the case, when the output buffer is larger
> than the input buffer, when you use effects like echo, delay, etc. The
> host would have to send an extra zeroed buffer, that has for example the
> size of the ringbuffer that has to be emptied. The host should be able to
> query that, if not the host could still sent zeroed buffers and check if
> they're coming back empty, but somehow I don't like the latter one.
>
I just joined the list so please flame me for jumping without knowing
what's happening. Is the plugin architecture geared for realtime effects
processing (replacing those little black boxes)? If this is the case it
seems easier to send same size vectors of samples between all plugins.
Because there is a constant stream of data you just wait until you receive
next vector and write the echo, delay, whatever there (or into whichever
vector you want). If you want to process fixed length audio files that is
another problem - still, you probably want all the plugins in a filter
network (I call this a dsp graph - what is the preferred term here?) to
process same sized arrays and not add length to the them and let something
upstream add the extra space to the file (your first suggestion). This is
especially true since empty (ie buffers with zeroes) may be valid after
processing - not likely but possible.
Karl
_____________________________________________________
| Karl W. MacMillan |
| Peabody Institute of the Johns Hopkins University |
| Network and Telecommunications Services |
| karlmac_AT_peabody.jhu.edu |
| 410/659-8297 |
-----------------------------------------------------
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:23:28 EST