This work presents a procedure for the estimation of a two-mass vocal fold model starting from a time-varying target „²ow signal. The model is speci„±ed by a large number of physical parameters, computed as functions of four articulatory parameters (three laryngeal muscle activations and subglottal pressure). Flow waveforms synthesized by the model are characterized by means of a set of typical voice source quanti„±cation acoustic parameters. Given a sequences of target acoustic parameters, dynamic programming techniques and interpolation based on Radial Basis Function Networks are used to derive sequences of articulatory parameters that lead to resynthesis of the target signal.

Index Terms. Voice source, Low-dimensional models, Estimation, Synthesis