On 2022-09-05, Peter P. wrote:
Andy, I wonder if your reverberated signal has a DC offset if only
positive samples are used. After all, the kernel would have one...
Since a convolution relies on both the impulse response of the filter
and the incoming signal, if you convolve with an impulse response having
a DC offset, the result will be entirely dependent on whether the input
signal *too* has a DC offset. I.e. a 0Hz component.
The easiest way to see the end result is to go to the Fourier domain,
and to apply the continuous time Fourier transform. That transforms a
convolution to a complex multiplication. That simplification even works
at zero hertz, i.e. at direct current. The formalism is entirely linear,
so that no differentiation between positive and negative samples is
needed.
Result to be told short, if your signal or your impulse response has a
DC offset, i.e. a 0Hz component, the result of a convolution will have
some as well. If either of then doesn't, neither will the outcome. So as
you can only control the reverberator's impulse response, and not its
input, make sure your impulse responses go to zero as their Fourier
transforms approach DC. Also do this gently, because as you
apodize/window either of time response or the reciprodical frequency
one, you'd be apt to introduce irregularities and oscillations on the
other side.
--
Sampo Syreeni, aka decoy - de...@iki.fi, https://urldefense.proofpoint.com/v2/url?u=http-3A__decoy.iki.fi_front&d=DwIBAg&c=009klHSCxuh5AI1vNQzSO0KGjl4nbi2Q0M1QLJX9BeE&r=TRvFbpof3kTa2q5hdjI2hccynPix7hNL2n0I6DmlDy0&m=DTxKzR-neLe0Sezw7HU7SJHnWnDI1-9bZU3busHdmka_gTXLxIHjNsqSztfb-xY0&s=yC6PRbWFbhfgHL5poTHGIAyf3pqM4DX8rvlXYoWPekM&e=
+358-40-3751464, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2