[Mp4-tech] Difficulties understanding MPEG-4 audio profiles andlevels

[Marc Jaeckle]

> The bad aspect of PCU and RCU is that they are imprecise. Max number of
> instantiations, channels, and sampling rates would be much easier to
> interpret, but would result in less flexability.
That's basically where my problem lies. I don't understand how the 
sampling rate influences the complexity (and therefore the RCU and PCU) 
of an audio object.  For some objects (e.g. AAC LQ or CELP) the PCU 
seems to be proportional to the  sampling rate but what about the RCU? 
Is there a way to calculate the RCU and PCU for an audio object at a 
given sampling rate or does the standard only give me the PCU and RCU 
for a single specific sampling rate for each audio object (like in the 
table "complexity estimates for the different audio object types")? Or 
to put it more general: How are the RCU and PCU calculated?
> ... but would result in less flexability.
I'm aware of the better flexibility that is achieved by describing 
levels with PCU and RCU that's why I was wondering why not all levels 
are defined by PCU and RCU. The first three levels of the Scalable 
Profile are defined by configuration for example and the Speech profile 
defines levels in terms of the max number of objects. I find this a 
little bit confusing and inconsitent. Was there a certain reason for 
this? (Well I guess there was a reason but what kind of reason :-)
Gruß
Marc