Quantum Signal Processing (QSP), a framework for implementing block-encodings of polynomial transformations of unitary operators, has become a cornerstone of optimal Hamiltonian simulation and related applications. Although QSP has been extensively studied, some limitations remain, including the angle-finding problem—that is, determining the sequence of phases required to implement a given polynomial—and extensions to the multivariate setting. In this talk, I will present a new perspective on QSP based on analysing the sequence of polynomials generated at each iteration of the algorithm. This viewpoint reveals a natural connection between QSP and the theory of orthogonal and biorthogonal polynomials, leading to new analytical formulas for the angle-finding problem. I will also discuss applications of this framework to multivariate generalizations of QSP, as well as to the implementation of polynomial series expansions, such as Hermite series.