Stuee,
You cannot simply apply power and audio to the device as the project's author implies. If you do, then the audio will cause the voltage of the resistor element to swing up and down, exceeding the absolute maximum negative voltage rating of the element (-0.5V), which will cause severe audio clipping and may damage the device.
In the author's Note #4, he advises to use a +,- 5V supply to solve this problem, but that is also an error since it would exceed the absolute maximum supply voltage of 8V for the device. He also failed to mention attenuating the input so that it doesn't exceed its rating.
That is why I mentioned biasing the pot and cap-coupling one end to ground when using a +8V supply, or using a +,- 4V supply.
It is difficult to make a quiet +,- 4V supply from a car battery, but it is easy to make a +8V supply from an LM317, then bias and cap-couple the element.
I must correct my statement about biasing and cap-coupling:
1) Bias the LOW end of the element to +4V, by using an equal-valued-resistors voltage divider of the supply voltage. Capacitive-couple the +4V to ground, so that audio is completely filtered out.
2) Attenuate the audio source so that it does not exceed 8V p-p.
3) Capacitive-couple the attenuated audio source to the HIGH end of the element.
4) Capacitive-couple the element's slider to your amp's input, or make certain that your amp has an input capacitor.
If this element is the same value as the existing volume control, then when paralleled the combination will double the source's low cutoff frequency.
How do you feel about the step sizes changing with volume setting, which makes the steps crammed together at the high end, and spread out at the low end?
Some day, maybe someone will make an audio digital pot.