I've seen circuit diagrams that use a polarize electrolytic to remove DC offsets from an input audio signal.
The cap is polarized. Why is this ok?
As long as the DC level is positive, the AC amplitude of the audio signal is of no consequence because it is a "coupling" capacitor, in the sense that the RC time constant is much higher than the highest 1/f of the audio signal (excluding dc of course), so the capacitor never gets "charged up" to the audio ac signal voltage (only to its dc value). Instead, the voltage across its terminals remains approximately constant. In other words, the capacitor rides with the input, and if the RC constant was not high enough, audio would be distorted. For example, you could have 1V of dc offset and 10V of ac amplitude. If RC is high enough, the voltage accross the capacitor will ALWAYS be very near 1V,NOT 11V to -9V as the comment to the first answer seemed to imply. Hope this helps.
Why wouldn't it be ok? In the usual configuration, the DC voltage on the positive side of the cap is always higher than the DC voltage on the negative side.
Even if you connect it with no bias, electrolytics can still withstand a small reverse voltage. It's only after they are reversed continuously at 1 V or more that the dielectric starts to degrade. See http://en.wikipedia.org/wiki/Electrolytic_capacitor#Polarity http://en.wikipedia.org/wiki/Capacitive_coupling