The relation between tongue shape and pitch in clarinet playing using ultrasound measurements

Abstract

Tongue shapes during clarinet performances of chromatic scale, portamento and pitch bending exercises were imaged using an ultrasound machine while audio and video were recorded synchronously. Analysis of the data from four participants revealed that tongue position varies non-monotically with note frequency in the lowest register (up to ≈ 440 Hz) and then descends monotonically as note frequencies rise through the upper registers. The descent of the tongue results in an expansion of the vocal tract's posterior oral cavity, resulting in a vocal tract input impedance maximum tuned to high frequency. Portamento and pitch bending were found to involve a raising of the tongue and concomitant contraction of the posterior oral cavity. The degree of contraction was similar in both portamento and pitch bending, suggesting a common acoustic mechanism.

Control of the vocal tract when experienced saxophonists perform normal notes and overtones

“…During the measurements, the participants could play the instrument with their desired volume in contrast to acoustic measurement methods [6,7,9,10], for which the participant should mime or play softly to obtain a high signal-to-noise ratio. While an ultrasound method can visualize the dynamic change in the tongue contour [21], MRI is also capable of imaging the vocal tract at a frame rate of several dozen frames per second [22]. For our future studies, we intend to examine a larger number of participants and wider range of vocal-tract shapes for various musical notes, to confirm and generalize the present findings.…”

Articulatory-acoustic relations in the production of alveolar and palatal lateral sounds in Brazilian Portuguese

“…Palate position was verified across several ultrasound image frames during the swallow and during linguo-palatal contact of consonants throughout the recordings. As reported in Lulich et al (2017b), the expected error for palate registration is no more than 2.5 mm.…”

“…The creation of area functions from 3D/4D ultrasound images required us to manually register the digitized palate impressions with the ultrasound images, and to segment the tongue surface in three dimensions. Errors associated with palate registration and tongue surface segmentation have been reported to be less than 2.5 mm (for registration; Lulich et al, 2017b) and less than the size of a pixel (for segmentation), which in the present study was less than 0.48 mm (Charles and Lulich, 2018). Although manual registration and segmentation is currently the gold standard for ultrasound analyses of speech sound articulations, they are onerous and would benefit from further development of computational tools that automate, or semi-automate, these processes.…”

Vocal tract shape and acoustic adjustments of children during phonation into narrow flow-resistant tubes

“…The median frame rate was 19.25 fps (Table I), and ultrasound intensity was recorded with a depth of 8 bits. Technical characteristics of the ultrasound system and similar studies using this system are described in Charles and Lulich (2018), Lulich et al (2018), andLulich et al (2017).…”