Tim Canfer

I am trying to get a reliable reading for dB of audio at each beat. Currently I am using an [average~ 1000 rms] object, a [snapshot 4n] object and an [atodb]. I had hoped that using overdrive would keep the signal relatively accurate, but unfortunately not. The simplified part of the patch is below:

and the dump for the first several bars of the same recorded audio is below:

Any suggestions on how to make this more constant?

print: -inf 0.
print: -29.902485 0.03198
print: -30.113321 0.031213
print: -35.468758 0.016849
print: -32.614075 0.023404
print: -26.64386 0.046538
print: -34.466808 0.018909
print: -35.575901 0.016642
print: -33.100227 0.02213
print: -31.480457 0.026667

print: -inf 0.
print: -30.956564 0.028325
print: -27.363419 0.042838
print: -37.287838 0.013665
print: -33.483303 0.021176
print: -28.729519 0.036604
print: -38.42881 0.011983
print: -36.480457 0.014996
print: -35.351788 0.017077
print: -31.982824 0.025169

wildly varying volume for same audio - any fixes?