Ali is no doubt correct that this may vary with accent, but some googling indicates that the general rule (which most native speakers will probably not be consciously aware of) is that the pluralizing -s is pronounced voiced, /z/, if the word ends with a vowel or glide, a voiced consonant, or a sibilant (in which case you have to insert a vowel so that you get -es, pronounced /ɪz/ or /əz/).
So you have (in half-assed notation):
tree-z (vowel)
eye-z (glide)
bed-z (voiced consonant)
wish-ez (sibilant)
However, when you have a word ending in an unvoiced consonant, the -s is unvoiced, /s/:
pin-s
cop-s
I'd also think, though, that the actual realization of these sounds will depend on the surrounding speech segments: for example, it's fairly common for sentence-final sounds to become spontaneously devoiced (because it's harder to articulate the voicing when there's no vowel following it).