Speaker- and Text-Independent Estimation of Articulatory Movements and Phoneme Alignments from Speech