[A6] (OT) neuron

[Brian Offermann]

>> Do the quality of neural models degrade as you use longer samples?
>
> Nope.

I'm not sure I asked that correctly...

My understanding is that it constructs a neural net that represents the 
spectral analysis of a sound over time.  This is a smaller file than 
the original.  And you can play this model as if it were the recorded 
sample, and it will sound like the original.

Say have two models:
A single electronic drum kick.  perhaps a second long.
B couple measures of a symphonic work.  much longer timeframe.

How do the models compare to the original samples.  I'm assuming Model 
A would be nearly indistinguishable from Sample A, where Model B would 
be a vaguely reminiscent approximation of Sample B?  Am I wrong?


This Neuron is an interesting contraption, no doubt!
I will shut up and download the manual now ;-)