[en] We define an asymptotically normal wavelet-based strongly consistent
estimator for the Hurst parameter of any Hermite processes. This estimator is
obtained by considering a modified wavelet variation in which coefficients are
wisely chosen to be, up to negligeable remainders, independent. We use
Stein-Malliavin calculus to prove that this wavelet variation satisfies a
multidimensional Central Limit Theorem, with an explicit bound for the
Wasserstein distance.