Stuee:
Initially, I was thinking of longer time delays than 5 and 30, so I just picked a few things:
A timer operating in mode #2 **broken link removed** followed by a one shot. e.g.
**broken link removed** These are probably "poor choices"
Hear me out for a moment: Don;t worry about possibe inversion for now.
Just to make it easier to understand, suppose we have an average audio level as a voltage.
We would then create a setpoint with hysteresis for the level we want to trigger at.
Nearly the same as the bar graph display, but not exactly.
You have a free-running oscillator that is "gated" by the set point.
So, below the threshold, the oscillator gets through.
Above the threshold, the oscillator stops.
This is then the trigger signal for a "re-trigerable monostable" with your 5 second time.
So, if it keeps getting triggered it stays, say OFF.
One it is triggered, you need to feed that into a "non-retrigerable timer. A timer that doesn;t reset on a new trigger.
Now this timer changes state instantly and keeps the audio off for 30 seconds.
It's VERY possible that with Liner's TimerBlox chips
http://www.linear.com/product/LTC6993-1 (not necessarily just this chip), you can basically achieve your goals.
www.jaycar.com is an .au outfit. You choice might be Farnell or RS electronics.
==
The point is, think about how it could work. A gated oscillator (square wave) + a retrigerable timer followed by an edge triggered non-retrigerable timer.