Looking at the efficiency curves from some T-amp chips, it seems that using the parallel-resistor trick might not result in a substantial increase in power consumption. The efficiency curves look like "constant idle power with nearly all the dynamic power delivered to the load". If we assume that the average output power is rather less than 1 watt - as would be the case for high-efficiency speakers at moderate listening levels (say 85 dB), then the "wasted" power is maybe a watt across both channels.
So, while it would be "nicer" to re-tune the output filter, I'm having a hard time coming up with an serious downside to the 16-ohm shunt resistor hack.
So, while it would be "nicer" to re-tune the output filter, I'm having a hard time coming up with an serious downside to the 16-ohm shunt resistor hack.