[microsound] registering volume modulation

[Justin Glenn Smith]

Renato Fabbri wrote:
> Hi Microsound List,
> 
> I am trying to learn what is the common way to register an audio
> waveform's intensity or volume modulation. As far as I can guess, the
> main issues are:
> 
> 1) What are reasonable chunk sizes (in miliseconds i assume)? And what
> overlapping fraction is used?
> and
> 2) What kind of measures are used for this chunks. (RMS/Peak)
> 
> I found nothing. It would be great to know about it in a general case,
> but the specific case is for voice analysis.
> 
> Does anyone knows how is this done or where can i find this?
> 

Most people cannot hear frequencies below 20hz, so any harmonic component
lower than 20hz is heard as volume change rather than a part of the timbre.

1/20*1000 gives 50 --- 50 ms or longer chunks should probably work. RMS 
will tell you more about the sound's perceived amplitude than peak will.
For extra credit you could do an fft and figure out the sone level for
the sound (the sone scale is weighted for human perceptual nonlinearities -
the frequencies that we hear best add more to the score).