Okay, here's the way this works:
For any "signal crosstalk" to occur between one set of conductors and another (two different source/return paths), there has to be either capacitive or inductive coupling. And the only way to generate capacitance or inductance between these cable pairs is if they route parallel to each other for a significant distance. (All capacitive/inductive coupling behaves via laws of electromagnetic fields - Maxwell's equations.)
For capacitive coupling to have a significant effect, both voltage and frequency have to be very high - that's the way capacitors work. For inductive coupling, the current has to be very high - since this is how inductors work. For car audio systems, voltage is relatively small, and frequencies don't get much above 20kHz (at least for audible affects). Thus, the capacitive effects from cable routing are very minimal.
DC power cable routing might have some influence on line level signal cabling, if there is any AC ripple in the amplifier power supply cabling. But, this can be minimized with proper grounding techniques, which ensure that RCA cable shielding will act to reject any inductive coupling that could occur. And this can be minimized further by maximizing cable separation as much as possible for parallel routing, and crossing cables at 90-degree angles. (Someone's already mentioned this - the coupling doesn't happen at 90-degree crossing angles; this from the way Maxwell's equations are stated.)
Bottom line - this isn't usually a big problem for car audio systems, at least as far as cable routing is concerned. But, noise can be injected into components from AC ripples in the car's DC supply system. This usually happens with lower quality head units, amplifiers, or other components. Rejection of this noise requires either higher quality equipment, or filtering on the DC power inputs. (another topic entirely....)
Hope this helps,
John