Well, yes I did. But that circuit we discussed relies on the fact that the black wire will be the reference point for GND.
So If I build two circuits I'd also have to build two power supplies in that setup.
The black wire would be the reference for GND if the outputs from your amplifier are ground-referenced.
Building two circuits does not require two power supplies.
When you measure the resistance between each of the black terminals (left and right) and chassis ground, be sure to measure it BOTH ways (exchange the leads on your meter). That way you can be sure that you have a hard connection vs. measuring through some semiconductor and giving a false indication.
I am not sure if I understood this correctly. I want a "Left GND Right" signal, but this way I'd get a mono signal or not?
You can use the resistors to form a "summing circuit" to mix the Left Speaker audio and the Right Speaker audio into a single L+R signal.
How exactly would I do that?
You have identified the differential input circuit that I was suggesting.
I would use these across the speaker wires, correct? Meaning that black and red drive the primary coils of the transformers. Then I would connect the secondary coils at one point and would obtain a common-ground, correct?
Yes, that is correct. And the "hot" side of the secondary winding of each transformer can be connected to a pair of resistors to sum the audio together into a monaural L+R signal.
Could you explain this in detail? I did not understand it completely. The "In a bridged set up, the other output is out of phase so as far as the meter is concerned it doesn't add additional information - it adds more amplitude." part I understood, since in a bridged output black would carry the inversion of red. But where do I get the ground? I could at a voltage divider between black and red of two 1K resistors and the middle would be ground. This could work in a bridged setup but obviously not in a non-bridged output (which is no problem, I could add a switch for that or something). Is this what you meant?
Yes, if you have two balanced, differential signals, then you don't really have a ground reference. But if you use differential input circuits, then you don't need a ground reference from the source, you create your own local ground reference.
After googling "differential circuit" I made this. The amplitudes are arbitrary and just for the sake of testing.
Would this work or is this bullshit?
I'd feed the outputs "left" and "right" into two builts of dannyf's original circuit.
The first is for a bridged, the second for a non bridged output.
Yes, that is the kind of circuit we are talking about. That is the active-circuit equivalent of using isolation transformers.
You don't really need R1/R2/R3/R4 if we assume you have a conventional audio power amplifier, whether differential or ground-referenced.
If the amplifier is ground-referenced, then the "-" side of "V2" and "V3" are ground by definition, and "V1" doesn't exist.
If the amplifier is bridged/differential, then the center points of V1/V2 and V3/V4 are ground, and "V5" doesn't exist.
The two circuits are essentially identical. So if you use a differential input, it will work with either type of amplifier.
If the amplifier is bridged/differential, then you have no external access to the amplifier's "ground" reference.