[LAU] [Zynaddsubfx-user] yoshimi bug fixes

fons at kokkinizita.net fons at kokkinizita.net
Sat Oct 24 06:51:59 EDT 2009


On Sat, Oct 24, 2009 at 05:51:19PM +1300, Chris Edwards wrote:

> I used to be put off by meters that looked slow,
> until I realised how useful RMS meters were and
> how long the windowing time needs to be to capture
> the bass frequencies. :)  Fons, I've been wondering,
> though, does the update rate in jkmeter affect the
> measurement window, or just the display rate?

It just determines how smoothly the display will appear.
The dynamic behaviour of jkmeters (and also of the VU and
PPM meters in jmeters) is very strictly defined. The values
that are displayed are calculated in a way that does not
depend on the display update rate, and (within reasonable 
limits) also not on the Jack period size. A level meter
would just be useless eye-candy otherwise.

There is indeed a 'window' for the RMS value, but
don't imagine it as being rectangular, i.e. with 
all samples having the same weight. The window is
more like an impulse with a slightly smoothed front
(which determines how it will react to short bursts)
and an exponential decay. It is the impulse response
of a critically damped 2nd order filter acting on the
square of the signal. The controlled fallback of
the RMS indication is not artificial (as it is for
the peak value), it *is* the shape of the window.
 
> I've also been wondering what the purpose of the
> fall-of on peak meters is.  Why not just reset the
> peak level immediately and let it get bumped up
> again by the signal?

Even in that case you would need a defined fallback
rate. Lower peak levels should be shown only if they
are not immediately following a much higher peak,
otherwise the peak level will jump up and down all
the time. In other words the bump-up should happen
only if the current input is higher than the fallback
of the previous peak. Having a visible fallback is
easier on the eyes.

Ciao,

-- 
FA



More information about the Linux-audio-user mailing list