Although the formalization of probabilistic programs already has several applications in the fields of security proofs and artificial intelligence, formal verification experiments are still underway to support the many features of probabilistic programming. We report on the formalization in the Coq proof assistant of a syntax and a denotational semantics a probabilistic programming language with sampling, scoring, and normalization. We use dependent types in a crucial way since our syntax is intrinsically-typed and since the semantic values are essentially dependent records. Thanks to the features of Coq, we can use notations that hide the details of type inference when writing examples. The resulting formalization is usable to reason about simple probabilistic programs.