thinking about the bridged tee attenuator that I designed (see earlier post), using break before make switches could be bad, as it occurs to me that when the series arm opens, that's the condition for infinite loss... but it's only infinite loss if the shunt arm gets to ground. Since (when the switch is operated), the series arm goes to open, and the shunt arm goes to open, then there's two resistors equal to the design impedance connected between input and output.
that could be loud, as that is not quite zero attenuation. Build one and see. You don't need to go nuts and buy 1% resistors, use nearest 5%. The attenuation values won't be quite right, and the impedance of the pad won't be constant, but it'll be close enough to confirm my suspicions.