Alternator whine in car (and airplane) audio is caused by a classic ground loop.
The car body is used as the return path for every power consuming appliance in the car, because each appliance is "grounded" locally to the car body. In some parts of the car, the current that is flowing through the car body adds up to several amps. The steel from which the car is fabricated is not a perfect conductor, so several amps of current in the body parts causes voltage drops of tens of mV.
Now, say you have an audio source like an MP3 player powered from a cigar lighter plug. It is connected to an audio amp mounted under the seat, which happens to be "grounded" to a screw to the floor under seat. A shielded cable with an 1/8" plug connects between the MP3 player and the stereo amp.
Most likely, the sleeve of the stereo jack on the MP3 player is internally tied to the negative wire of the cigar plug powering the MP3. Most likely the shield on the cable running to the audio amp is tied to the ground inside the amplifier, maybe through the shell of a RCA phono jack. The power ground wire from the amplifier is internally tied to the same PC trace as where the shell of the RCA jack, and as we have already said, it is connected to a screw under the seat...
So, the MP3 is "grounded" via the cigar plug. The amplifier is "grounded" to the screw under the seat. We get alternator whine. How is it getting into the system?
As I said above, there are sizable voltage drops between various parts of the car body. Most of the time, the alternator carries the total electrical load. It uses a three phase full-wave rectifier system which puts out mostly DC, but there is about 10% ripple at an integer multiple of the rotation rate of the alternator, and it happens to be plainly audible (about 1-3kHz). The ripple rides on all currents flowing to and from various appliances in the car.
So suppose that there is 30mV of alternator ripple between where the cigar plug is "grounded" to the car body (under the dash), and the part of the car under the seat where the amp is "grounded". That 30mV will appear effectively to be in series (common-mode signal) with the audio output from the MP3 player.
But wait a minute, isn't the output audio from the MP3 at "line level", something like 2Vp-p? 30mV compared to 2V shouldn't be a problem, however, 20log(.03/2) is only -36db, and what is the range of human hearing?
No amount of "filtering" with capacitors (even inductors) will cure this. There are only a couple of things you can do. First, suppose instead of operating the MP3 player using a cigar plug power cord, you ran it on its internal batteries. If there is no connection to the car body at the MP3 player, and the only connection form it to the amplifer is via the shielded audio cable, then that might cure the problem.
If you powered the MP3 from inside the amplifier by back-feeding DC power from amp to MP3 while at the same time feeding the audio from MP3 to the amp, you have created a "single point ground" locally near the amplifier, where no part of the audio system touches the car body anywhere else in the car except at this single point, and that usually works. Bear in mind, that this is sometimes hard to do, because maybe the "amp" is really a stereo receiver, which also has an AM-FM antenna mounted elsewhere on the car body, so is already part of a different ground-loop.
Another strategy to break ground-loops involves transformer coupling. Suppose the MP3 player is powered via the cigar plug. There is a large common-mode ground-drop between the cigar plug ground and the amp ground. If you insert an audio coupling transformer in the audio path between MP3 and amp, the isolation from transformer primary to secondary effectively breaks the ground-loop.
I know this game. I have spend years of my life tracking down ground-loops in industrial and robotics instrumentation, and more recently getting alternator whine out of aircraft audio entertainment systems.