Even for Latin, there are several ways to pronounce it and we don't know exactly how it was pronounced. Reconstructed PIE is less certain, so the phonetics is vague. In terms of the orthography representing contrastive sounds, you can make a fair approximation, which might vary based on certain theoretical details (for example, whether you are using glottalic theory or not). But overall it should be fairly similar to what is expected from phonetic transcriptions in general. As far as I know almost all letters have their IPA values.
This page has some relevant information:
http://en.wikipedia.org/wiki/Proto-Indo-European_phonologyI'd assume that the ĝ you have is the palatal voiced stop (see the wiki page), which might have been pronounced as a front [g] (as in 'gill', not 'go') or as an affricate as in modern English 'judge'. I'm not sure exactly what is said about that detail.
For h1 (and h2 and h3), see:
http://en.wikipedia.org/wiki/Laryngeal_theoryThe sequences <bh>, <dh>, etc., represent voiced aspirated sounds (as in modern Hindi, or earlier in Sanskrit), but I believe there's some controversy over exactly how they were produced. You might want to take a look at this page as well:
http://en.wikipedia.org/wiki/Glottalic_theoryThe rest are their standard IPA values, I believe.
Remember, these reconstructed phonemes are meant to be the origin of sounds found in modern languages, so they are more accurately connected to those (at least some of them) than they are really representations of how PIE was pronounced. So in addition to learning how to pronounce these reconstructed words, you should also pay attention to the sounds they relate to in the modern languages. So <p> was pronounced roughly as /p/, and is preserved in many daughter languages, but in Germanic it is found as /f/ (see Grimm's Law). So while knowing that <p> is pronounced [p] in that reconstruction is somewhat important, it is at least as important to know that it relates to /p/ in most daughters and /f/ in Germanic.
