I am reading about variational auto encoders  and there is the bellow loss function:  
$li(\Theta,\phi) = - {\mathbb{E}}_{z\sim q} [log p_{\phi}(x_i|z)] + KL(q_{\phi}(z_i|x)||p(z))$  

My question is: What does the notation $z\sim q$ under $\mathbb{E}$ mean. I just have seen for expected value $\mathbb{E}$ notations like $E(X)$ or $ \langle X\rangle $.  
Could some explain what this notation generally means when using  $\mathbb{E}_{x\sim y}$ for some $x$ and some $y$?