[LAU] LADSPA processing time for audio zeroes

Fons Adriaensen fons at kokkinizita.net
Sat Mar 28 13:56:26 EDT 2009

On Sat, Mar 28, 2009 at 05:09:12PM +0100, Nigel Henry wrote:

> Your not experiencing a denormals problem perhaps with the plugins? Googling 
> denormals bring up some interesting hits on the first page. May be worth a 
> look if that is the problem.

In general, zero input is a signal just like any other,
processing should go on. It does not automatically imply
that the output should be zero. Filters will continue to
produce non-zero output after a signal has stopped, for
some time. An IIR filter will decay to zero exponentially.
This is usually also the point where denormals will be

Some plugins could be made to simplifiy their processing
when given zero inputs for a long enough time. The logic
required to do this could be worse than the problem it
tries to solve.

A similar problem exists with DSP code that interpolates
its control parameters to avoid clicks or zipper noise:
the logic required to optimise the constant control value
case can become quite complicated in some cases.

The only place where avoiding zero inputs is likely to
pay off is a polyphonic synth patch, but even here it
would require some logic - notes usually don't end on
a MIDI note-off. When to stop a particular voice would
depend very much on the patch details.



Io lo dico sempre: l'Italia è troppo stretta e lunga.

More information about the Linux-audio-user mailing list