Acoustic Identification of the Voicing Boundary during Intervocalic Offsets and Onsets Based on Vocal Fold Vibratory Measures

Abstract

Methods for automating relative fundamental frequency (RFF)—an acoustic estimate of laryngeal tension—rely on manual identification of voiced/unvoiced boundaries from acoustic signals. This study determined the effect of incorporating features derived from vocal fold vibratory transitions for acoustic boundary detection. Simultaneous microphone and flexible nasendoscope recordings were collected from adults with typical voices (N = 69) and with voices characterized by excessive laryngeal tension (N = 53) producing voiced–unvoiced–voiced utterances. Acoustic features that coincided with vocal fold vibratory transitions were identified and incorporated into an automated RFF algorithm (“aRFF-APH”). Voiced/unvoiced boundary detection accuracy was compared between the aRFF-APH algorithm, a recently published version of the automated RFF algorithm (“aRFF-AP”), and gold-standard, manual RFF estimation. Chi-square tests were performed to characterize differences in boundary cycle identification accuracy among the three RFF estimation methods. Voiced/unvoiced boundary detection accuracy significantly differed by RFF estimation method for voicing offsets and onsets. Of 7721 productions, 76.0% of boundaries were accurately identified via the aRFF-APH algorithm, compared to 70.3% with the aRFF-AP algorithm and 20.4% with manual estimation. Incorporating acoustic features that corresponded with voiced/unvoiced boundaries led to improvements in boundary detection accuracy that surpassed the gold-standard method for calculating RFF.

A Scoping Literature Review of Relative Fundamental Frequency (RFF) in Individuals with and without Voice Disorders

“…Further exploration into improving the accuracy of algorithmically extracted RFF values has yielded a new set of acoustic feature processing techniques to locate the voicing cycles closest to the voiceless consonant (i.e., offset cycle 10, onset cycle 1). This most recent version of the RFF algorithm (aRFF-APH) [52] was built by integrating information from simultaneous recordings made using a microphone and high-speed flexible laryngoscopy. The authors were able to identify the physiological-rather than acoustic-initiation and termination of vocal fold vibration that marks the boundaries of the voiceless consonant.…”

“…Five studies focused on algorithmic development for automated RFF extraction [33,39,40,51,52]. Semi-automated algorithmic development began in 2013, in which a custom MATLAB (The MathWorks, Natick, MA, USA) program extracted RFF values from VCV utterances recorded with a microphone [40].…”

“…The algorithm (aRFF) [39] was updated in 2017 and openly shared on Boston University's website at no cost to the user. Vojtech and colleagues [51] refined the algorithm via incorporating rule-specific criteria for tuning algorithmic parameters based on patient-specific voice characteristics and acoustic recording environment (referred to as aRFF-AP; current version) and, later, by cross-referencing simultaneous high-speed laryngeal imaging to hone in on cycle-specific extraction decisions around the voiceless consonant (referred to as aRFF-APH; unreleased) [52].…”

Segmentation of Glottal Images from High-Speed Videoendoscopy Optimized by Synchronous Acoustic Recordings

“…Other very recent work concerns the determination of the effect of incorporating features derived from vocal fold vibration transitions into acoustic boundary detection [37], comparative analysis of rapid videolaryngoscopy images and sound data [38], and a computer model for the study of unilateral vocal fold paralysis [39]. Interestingly, a method for detecting COVID-19 by analyzing vocal fold vibrations has also been proposed [40].…”

Special Issue on Computational Methods and Engineering Solutions to Voice II

“…Schlegel et al [15] perform investigations of acoustic parameters using in vivo porcine models to enable quantitative voice outcome tracking of laryngeal surgical interventions for porcine models. Vojtech et al [16] analyse phonatory offset and onset processes and characteristics, using the relative fundamental frequency.…”