Articulatory Controllable Speech Modification using Sequential Inversion and Production Mapping with Gaussian Mixture Models