[PD] Real-time frequency filtering and analysis

[Jamie Bullock]

Hi Jared,


On Tue, 2007-03-27 at 23:20 -0500, Jared wrote:
> Summary: I'm interested in capturing audio through a microphone and, in 
> real time, determining which frequency range (from a set range of 
> frequencies) has the maximum level. This would be used to determine 
> which note is loudest out three notes.
> 
[snip]
> For the voting, I plan for everyone in the audience to have a small 
> xylophone with 3. Each note will correspond to a voting choice presented 
> by the cast. I'm going to have microphones positioned around the theatre 
> to capture the xylophones and pipe them to a computer. What I'd then 
> like to do through PD is determine which note was 'loudest,' meaning the 
> most people cast it for their vote.
> 

You could try using a multiband filter and some envelope followers, but
I don't think you will be able to measure the voting very accurately
using this technique. 

To get slightly better results, you might want to take into account the
effect of frequency on perceptual loudness (cf. Zwicker et al). You
should also take into account whether or not the various votes are
synchronised in time by taking measurements over some time frame, and
averaging the results.

There are also other issues to consider such as the fact that not all of
the voters will strike the xylophone equally hard, the
resonance/absorption qualities of the room, and establishing the correct
relative contribution from each microphone.

Basically, this is a difficult problem to solve, but if all you need is
a 'rough idea' of how many votes have been cast, you might be OK.

Jamie