As it is known, $\mathbb CP^n$ with Fubini-Study symplectic form can be get by the symplectic reduction of $\mathbb C^{n+1}$ with a symplectic form $\sum_{i=0}^n dz_i\wedge d\bar z_i$ by a hamiltonian action of $S^1$:

$$ t\cdot (z_0,...z_n) = (e^{\sqrt{-1}t}z_0,...e^{\sqrt{-1}t}z_n). $$ The resulting Fubini-Study form is equal (up to a coefficient) to $$ \omega = \partial\bar\partial \log\sum_{i=0}^{n}z_i\bar z_i $$

One can try to generalize this form to a weighted projective space $\mathbb P(\lambda_0,...\lambda_n)$ in such a way:

$$ \omega = \partial\bar\partial \log\sum_{i=0}^{n}z_i^{\mu_i}\bar z_i^{\mu_i}, $$ where $\mu_i$ are integers such that $\lambda_i \mu_i = const$.

The problem is that this form is not the form you get by symplectic reduction of $\mathbb C^{n+1}$ by the $S^1$ action

$$ t\cdot (z_0,...z_n) = (e^{\sqrt{-1}\lambda_0t}z_0,...e^{\sqrt{-1}\lambda_nt}z_n), $$

although the weighted projective space is exactly the result of this construction (of course, if you choose a non-zero element of the dual $S^1$ Lie algebra).

My question is: what is the right generalization of Fubini-Study form to weighted projective space? By the right generalization I mean the form resulting from symplectic reduction.