Okay, I think a break through: I tried again freezing ROM IC A more thorough, not only the top of the IC but also underneath it and meanwhile played a patch that normally sounds cracking. And from power-on the patch sounded clear for 8 to 10 key presses, where it normally starts cracking and breaking up after 2-4 key presses! So the freezing actually did improve the situation albeit very temporarily. I have repeated this a few times and the improvement is consistent with every thorough freeze.
So it looks like ROM IC A is faulty, do you agree?
(I've also tried soldering-through the IC-'legs' again, but without any improvement.)
Now I have found the ROM image files I think the solution would be either 'burning' a new ROM IC using the correct ROM image or try finding a replacement original ROM IC. Both options are not going to be easy...
However, just checked the pins of the original PCM ROM IC A and compared it with the pins of this 2mbit 32-pins EPROM and they seem to match, the M27C2001
https://pdf1.alldatasheet.com/datasheet-pdf/view/228430/STMICROELECTRONICS/M27C2001.html.
Would this EPROM be a good replacement for the original ROM IC?