If it were me trying to solve this, I would not bother with shielding wires as this is difficult and not entirely productive. I would start by doing some experiments to simplify the system in an attempt to isolate the mechanism by which the energy is coupling into the audio system. If the head unit is separate from the amp, one experiment might be to disconnect the head unit but leave the amp in place and then try another audio source with short wires directly to the amp or for that matter just leave the amp inputs disconnected at the amp and power the map up and see if the cellphone interference can still be heard in the speakers. You could also reverse this experiment by connecting the existing head unit to a separate amp. In each case you need to find a way to remove parts of the system or use alternate parts that are remote from the interfering cellphone. Sort of a "divide and conquer" approach.
If you find a line input, for example, that when removed makes a huge difference to the problem, then the next thing I would do would be to open up the amp or head unit and solder a small ceramic capacitor, (the smaller the better if leaded, or a chip type if possible), between the hot line and chassis or circuit board ground. To avoid corrupting the operation of the system, you have to keep the capacitance small, but it still has to be a large enough value to have an impedance of less than a couple of ohms at 850 MHz. Perhaps 27 pF.
Yes, ferrite beads may be a very good idea as I have used them before and they do indeed help. You have to use the right kind of ferrite beads since many of them are made to be most effective at lower frequencies.
This sort of thing:
https://www.fair-rite.com/cgibin/catalog.pgm?THEAPPL=Suppression+Components&THEWHERE=Cable+Component&THEPART=Round+Cable+Snap-Its#select:freq1
can work but you have to pick the type suited to the highest frequencies. You can also buy a kit of assorted shapes from this company and others too. Nice to have a kit for experimentation. If you have beads handy it is easier to try these before fooling around with soldering capacitors inside the boxes.
I have no doubt that your problem is rectification of the cellphone transmitter energy within the head unit or amp. GSM transmissions have sharp changes in amplitude at a rate of 220 Hz (approximately) and this (plus audio harmonics) is what you hear. The RF energy is a high enough frequency that it can couple very easily from one trace to another once it is inside the box so it tends to "flow" everywhere to varying degrees. Any PN junction that sees this RF voltage can rectify it and some are worse than others due to junction capacitance and lead inductance. There are many more PN junctions than you think inside equipment like this because every diode, every bipolar transistor, every JFET and many other semiconductors have PN junctions, and most ICs have protection diodes built in to their I/O pads that are also PN junctions. So there are lots of places for the rectification to occur.
If you can't stop the RF from coming in by treating cables and connections, the next thing to try is to apply bypass capacitors across the most likely PN junctions. Since a small value capacitor is very low impedance to such RF, putting one across a PN junction forces the voltage on both sides of the junction to be identical and with no potential difference across the junction there is no rectification. I mention this as a bit of background theory so that you can understand that the components used to solve the problem are cheap and simple, but finding where to put them is very difficult.
Fixes that are helpful are often accumulative. For example, a ferrite bead on one cable might not fix the problem and a ferrite bead on another cable might also not fix it, but both ferrite beads on both cables might indeed make a difference. So, as you go along, you might find it useful to leave some fixes in place as you try others.