A novel Statistical Approach for F0 Estimation, SAFE, is proposed to improve the accuracy of F0 tracking under both clean and additive noise conditions. Prominent Signal-to-Noise ...
This paper applies time-varying autoregressive (TVAR) models with stochastically evolving parameters to the problem of speech modeling and enhancement. The stochastic evolution mod...
Speech inversion is a way of estimating articulatory trajectories or vocal tract configurations from the acoustic speech signal. Traditionally, articulator flesh-point or pellet t...
Vikramjit Mitra, Hosung Nam, Carol Y. Espy-Wilson,...
Shannon's Noisy-Channel model, which describes how a corrupted message might be reconstructed, has been the corner stone for much work in statistical language and speech proc...