I learned the IPA entirely online. I don't see why its that much of an issue.
I started with learning how sounds are produced. The first things I learned were the concepts of 'place of articulation' and 'manner of articulation'. The first is obviously easier to explain.
You can pretty easily teach someone about place of articulation. For example, you can explain that p, b, and m are all bilabial consonants, because they're pronounced by touching the lips together. F and v are labio-dental, because they are pronounced by touching the upper teeth to the lower lip. Etc etc...
If you want an example of how it can be done, go look at the Language Construction Kit. Though admittedly, it might be somewhat difficult without being able to show someone a chart, but you can find plenty of charts online that you can point them to. Also, its not that hard to make a consonant chart in a word program.
As for learning new consonants? That's harder. I myself had no problem learning the voiceless velar fricative from a book, but that may just be me. Also, being an English-speaker, my native language has both velars and fricatives. Learning a velar fricative wasn't really that much of a leap. Though come to think of it, I was also able to learn the voiceless palatal fricative just as easily.
Normally, learning a sound when you can't hear what it sounds like is hard, unless you're already very familiar with how phones/phonemes are produced. Wikipedia has sound files for all the symbols of the IPA (each symbol has its own article), but the sound quality isn't the greatest. Either way, it is something.