5 Aug
2005
5 Aug
'05
11:41 a.m.
Interesting. So
again I'm inclined to think that the reported problem
is due to denormals.
I usually do a very blunt ... + 1e-20 at strategic places. So far,
nobody has complained about the DC offset :-)
Similar approach works fine here, too (flipping the sign of the
addition constant at every sample or every block, depending on the
algorithm in question).
Inaudible (in fact barely measurable), code is branchless and simple.
Perfect solution for a stupid little problem.