I suspect the residual noise from slewing is because its only a 1st-order correction to the previously discontinuous gain
changes - the derivative of the gain still has discontinuities at the start and ends of the ramped transitions, which contains
audible energy, albeit much less than before.
If you take a DC level as the gain setting, aggressively low pass filter this to below 20Hz, and multiply with the signal, you can
in theory remove any audible noise from gain changes - in practice this might add some small extra quantization noise
correlated with the gain changes. It involves quite a bit more processing though.
changes - the derivative of the gain still has discontinuities at the start and ends of the ramped transitions, which contains
audible energy, albeit much less than before.
If you take a DC level as the gain setting, aggressively low pass filter this to below 20Hz, and multiply with the signal, you can
in theory remove any audible noise from gain changes - in practice this might add some small extra quantization noise
correlated with the gain changes. It involves quite a bit more processing though.