CVS Commit 0.97.04

Fixed the problem with the RMS calculation. It was very simple in the  
end, though it took some finding.

In the approximation function we divide some positive number in the  
range [0,1] by the previous estimate, but if the previous estimate  
has been 0 then we get Inf as the value next time round, and that  
takes a while to decay back the the real value.

To get round it I just added a small bias to the input value, so it's  
never truly zero.

This will have been causing the CPU load in JAMin to go up when there  
was silence input - doing calculations on Inf is expensive - but it  
may not have been observable over the load from the FFT.

- Steve

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/