FIG. 1.

(Color online) Flowchart of signal processing procedures implemented in the Praat software to create a continuum of speech sounds varying in formant structure. Formant contours from /ba/ and /da/ were used as endpoints between which 6 other contours were interpolated to create FormantGrids. Each of those FormantGrids was used to filter a residual sound that was created by removing the formant peaks from the original /ba/ sound using LPC inverse filtering. After the filtering, each resulting sound was low-pass filtered and added with high-pass frequency energy from the original /ba/ sound.