Perceptual Evaluation of Vocal Fold Vibratory Asymmetry

Abstract

Objectives: Laryngeal vibratory asymmetry occurring with paresis may result in a perceptually normal or abnormal voice. The present study aims to determine the relationships between the degree of vibratory asymmetry, acoustic measures, and perception of sound stimuli.

Study design: Animal Model of Voice Production, Perceptual Analysis of Voice.

Methods: In an in vivo canine model of phonation, symmetric and asymmetric laryngeal vibration were obtained via graded unilateral recurrent laryngeal nerve (RLN) stimulation simulating near paralysis to full activation. Phonation was performed at various contralateral RLN and bilateral superior laryngeal nerve stimulation levels. Naïve listeners rated the perceptual quality of 182 unique phonatory samples using a visual sort-and-rate task. Cepstral peak prominence (CPP) was calculated for each phonatory condition. The relationships among vibratory symmetry, CPP, and perceptual ratings were evaluated.

Results: A significant relationship emerged between RLN stimulation and perceptual rating, such that sound samples from low RLN levels were preferred to those from high RLN levels. When symmetric vibration was achieved at mid-RLN stimulation, listeners preferred samples from symmetric vibration over those from asymmetric vibration. However, when symmetry was achieved at high RLN levels, a strained voice quality resulted that listeners dispreferred over asymmetric conditions at lower RLN levels. CPP did not have a linear relationship with perceptual ratings.

Conclusions: Laryngeal vibratory asymmetry produces variable perceptual differences in phonatory sound quality. Though CPP has been correlated with dysphonia in previous research, its complex relationship with quality limits its usefulness as clinical marker of voice quality perception.

Level of evidence: NA, basic science Laryngoscope, 131:2740-2746, 2021.

Keywords: Vibratory asymmetry; cepstral peak prominence; in vivo phonation; sort-and-rate; voice quality.

Fig. 1.

The visual sort-and-rate task as implemented in Microsoft PowerPoint. Listeners clicked each icon to play the associated voice sample. They subsequently dragged each icon to sort the stimuli from best to worst in the box provided below. All icons in the slide belong to the same set of graded RLN stimulation. RLN = recurrent laryngeal nerve.

Fig. 2.

Graded RLN level and neuromuscular compensation (bilateral SLN, contralateral RLN) interacted to determine symmetric or asymmetric vibration in both experiments. (A) In experiment 1, increasing RLN stimulation increased and then decreased the likelihood of symmetric vibration, depending on compensation condition. (B) In experiment 2, the likelihood of symmetric vibration increased with increasing RLN stimulation level, and the pattern of increase was determined by compensation condition. RLN = recurrent laryngeal nerve; SLN = superior laryngeal nerve.

Fig. 3.

Interaction between RLN level and neuromuscular compensation on perceptual rating in both experiments. Error bars represent 95% confidence intervals. (A) In experiment 1, sound samples from low RLN levels overall had high ratings, sound samples from high RLN levels overall had low ratings, and sound samples from levels 4 to 8 had varying ratings depending on contralateral compensation. (B) In experiment 2, this relationship was also observed, although compensation condition had more of an effect when comparing ratings from the same RLN level. RLN = recurrent laryngeal nerve.

Fig. 4.

Effects of vibratory symmetry/asymmetry on mean perceptual rating. Zero represents vocal fold symmetry and negative values represent incremental levels of asymmetry. Error bars represent 95% confidence intervals. (A) In experiment 1, symmetric conditions were associated with higher ratings overall. (B) In experiment 2, moderately asymmetric conditions (−4, −5, −6, −7, −8) were associated with superior ratings.

Fig. 5.

Relationship between CPP and perceptual rating in experiments 1 and 2. (A) In experiment 1, there was a weak correlation between CPP and perceptual rating; a CPP value of ≤10 represented a threshold for poor ratings. (B) In experiment 2, there was a negligible correlation between CPP and perceptual rating. CPP = cepstral peak prominence.