Nobody yet has mentioned the actual accidental amplifier ultimate power. All decent amplifiers are rated by the power level at some very low distortion level, like 0.1% total distortion. Sony car stereo systems were rated at 10% total harmonic distortion, while my old MARANTZ-2235 was rated at 0.5% THD. Of course the Sony car stereo systems would be about 1% THD at the 8 watt output, the high distortion was at the 40 watt level.
The CROWN DC150 was very low distortion at150 watts, but it could easily destroy 200 watt speaker systems when seriously over driven.
SPEAKER protection demands limiting the feed to a bit less than the specified power rating.THAT is why I suggest the diode clamp in both directions. Don't try to figure out triggering a triac, use power diodes that conduct when the voltage rises above 0.65 volts. (full conduction about 0.7 volts.) .
The biggest source of that power blast often is when somebody tries to plug in a mic or instrument when the system is already powered on and working at a high power setting . That momentary blast of noisy mains frequency sometimes is followed by silence of blown out speakers..

I presume you mean microphone inputs, but if they are phantom powered they will already have sufficient protection from the 48V phantom power. You can’t clamp line level inputs with a standard diode, because 0.775V rms or 1.228V for a +4dBm input would be too high.
And no clamping diode is going to protect a speaker from an amplifier that has gone DC, only a triac will do that. (a relay might, but not if its contacts got welded during the failure)

 

My first suggestion was ADDING THE CLAMPING DIODES TO THE FUSE! BOTH are needed. The diodes need to be POWER diodes, able to carry enough current to clear the fuse. Neither by itself is enough.