[LAU] JAMin updates?

Fons Adriaensen fons at kokkinizita.net
Tue Aug 4 17:40:50 EDT 2009

On Tue, Aug 04, 2009 at 08:57:55PM +0200, Arnold Krille wrote:

> I talked to the main-devs (was it steve harris) one or several times. As far 
> as I remember the 10ms delay is because of the math in either the EQ or the X-
> over for the compressors, don't remember exactly which. And there is probably 
> no way to cirumvent that.

It depends on two things. 

The first is the resolution of the EQ at low frequencies.
The GUI of Jamin's graphical EQ gives the impression that
you can set the gain at e.g. 20Hz and 25Hz independently.
This is not true and to make it possible you would need
a delay of around 1 / (25 -20) seconds, or 200 ms, at least
if you want linear phase. In the mininum phase case the
delay depends on the derivative of the frequency response,
so more differene between nearby frequencies implies more

This is more or less evident if you use an FFT (as the
difference between the frequency bins is 1 / length), but
it is really a general rule.  It's easy to see that detail
in the time domain (a waveform with short spikes in it)
requires a wide frequency range. The dual of this (swapping
the time and frequency domains) is that for detail in the
frequency domain you need more time.

The second reason to allow delay in a mastering app is that
the kind of agressive limiting and compression  expected from
such an app can be implemented *much* better if you allow the
processor to look ahead, which again implies delay.

In fact the best possible mastering process would be off-line,
non real time.

> Hmm, maybe DSP-expert Fons has something in his toolbox?

I have been considering writing something similar for a long
time, and during the past year I've been experimenting a lot
with dynamics processing (which will sooner or later result 
in some application). But as for most of us I've only got
25 hours in a day and 8 days per week.

One poster mentioned foh use. What functionality would be
required in that case ? Note that it's generally a bad
idea to drive a PA system to its limits.



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

More information about the Linux-audio-user mailing list