Fix for poorly designed class-D amplifier mute circuit

Question

I recently made a PCB for a project with several class-D amplifier channels to drive small speakers for audio feedback. I included a mute switch which connected or disconnected one end of these speakers from ground. However, because all the negative terminals of the speakers are connected to the mute switch, the audio still sees a path to ground through the other speakers and their amplifier circuits. I realized after the fact this was a poor design choice, pretty obviously.

I am now wondering if there's anything I can do as a workaround without getting new boards printed. For this project it's important that the visual quality of the board does not get ruined in the process, so I can only make changes to the back of the board or under the speaker components. Within those constraints I'm open to cutting traces or hot-wiring stuff on the back as needed. Please advise, any help is much appreciated.

I get this is a strange question but felt that it was worth a shot. This was an expensive PCB run and I would prefer not to have to reprint. Please no punching down or criticism of the original design as obviously flawed or saying "there is no solution make a new board" (I realize my original design is bad, that is not my question, and these responses are not helpful).

Here is a picture of my badly designed audio amplifier circuit and the section of the board for reference:

Answer 1

Here's a potential solution that I thought of. I can cut the trace connecting all the negative terminals of the speakers together, then solder two diodes with opposite polarity to each negative terminal, one connecting to the mute rail, and the other to ground:

^This circuit x 10, duplicated for all speakers

I was originally trying with a single diode in series, but this caused the class D amplifier to cease functioning since it disrupted the push-pull nature of the circuitry. With two opposite polarity diodes both going to ground the amplifier worked again, with the only cost being a very slight drop in amplitude. I realized that by grounding or ungrounding all the diodes of a given polarity I could accomplish the functionality of the mute circuit and also block any current from flowing back into the amplifiers.

Can anyone see a problem with this approach other than the slight volume drop? The sound is a square wave so distortion is not a concern.