In what order should I isolate and attenuate an audio signal

Detecting the signal - do you want to do it the hard way or the easy way - if the signal amplitude is in the order of 3 or 4 V peak-to-peak then it can be done the easy way using a couple of signal diodes, a couple of capacitors and a resistor. This would be fed via a Schmidtt trigger into an IO line. At a push the Schmidtt may be avoided if the logic levels are 3.3V.

The hard way means that the signals levels are lower than 3 or 4 volts peak-to-peak and an input amplifier needs to be inserted to boost the signal up to these sort of levels.

There are simpler audio detectors based around a couple of transistors and these might be suitable too but, due to their simplicity they may trigger on noise or crosstalk. These might be good enough though.

There are a few ICs that can apply large amounts of gain and then large amounts of compression to boost small signals to the same amplitude as larger signals and these could play a role.

None of the circuits described above will cause any appreciable deterioration of the original analogue signal but care would need to be taken when connections are made to audio lines.

The answers are all here (above) and to move further you need to do a little work on measuring the signal levels or opt for the more sophisticated system with gain so that small-ish signals can be detected and acted upon.

Soft limiting can be a really simple way of preserving the gain for smaller signals whilst limiting (with some measured distortion) the output volume for bigger signals. Feeding an input to a PA stage can be intercepted with back-to-back diodes so that for small signals the diodes are barely conducting whilst higher amplitudes get limited to about 1.4 Vp-p.

Another way is to use an AGC (automatic gain control) amplifier such as this one: -

Q1 is non-conducting when small signals are fed into the circuit. This means that Q2 (P channel JFET acting like a variable resistor) is biased into pinch-off by R8 and, doesn't attenuate the incoming signal (to ground) via R2. As the signal gets bigger Q1 starts to conduct and this drags the gate voltage on Q2 down towards ground turning on the JFET and attenuating the input signal. C4 acts to smooth Q1's effect and give more of a DC control signal to Q2 based on what the op-amp's output amplitude is.