An evaluation of rectilinear digital waveguide mesh in modelling branch tube for English nasal synthesis

This research presents the evaluation from applying a Two-Dimensional Rectilinear Digital Waveguide Mesh (2-D RDWM) to model the vocal and nasal tracts for synthesizing the three English nasal consonants /m/, /n/ and /ŋ/. The rectilinear meshing represents the propagated pressure according to the cross-sectional area at each tract position. Nasal tract coupling is achieved by modifying the number of ports from four to five at the location of the velopharyngeal mesh junction to attach nasal tract mesh. The proposed system exhibits zeros from the branch tube (closed-end oral cavity) and also nasal frequencies from the nasal tract. Spectrograms of the acoustic output for synthesized nasals exhibit zeros and nasal tract resonant frequencies comparable to those found in synthesized and real speech.