Harold A. Cheyne

Mobile Voice Health Monitoring Using a Wearable Accelerometer Sensor and a Smartphone Platform

Many common voice disorders are chronic or recurring conditions that are likely to result from faulty and/or abusive patterns of vocal behavior, referred to generically as vocal hyperfunction. An ongoing goal in clinical voice assessment is the development and use of noninvasively derived measures to quantify and track the daily status of vocal hyperfunction so that the diagnosis and treatment of such behaviorally based voice disorders can be improved. This paper reports on the development of a new, versatile, and cost-effective clinical tool for mobile voice monitoring that acquires the high-bandwidth signal from an accelerometer sensor placed on the neck skin above the collarbone. Using a smartphone as the data acquisition platform, the prototype device provides a user-friendly interface for voice use monitoring, daily sensor calibration, and periodic alert capabilities. Pilot data are reported from three vocally normal speakers and three subjects with voice disorders to demonstrate the potential of the device to yield standard measures of fundamental frequency and sound pressure level and model-based glottal airflow properties. The smartphone-based platform enables future clinical studies for the identification of the best set of measures for differentiating between normal and hyperfunctional patterns of voice use.

Ambulatory monitoring of disordered voices

Objectives: Recently developed systems for ambulatory monitoring of voice use employ miniature accelerometers placed at the base of the anterior neck to sense phonation. As it is hoped that such systems will help improve the clinical assessment and management of voice disorders, this study was undertaken to determine the impact of dysphonia severity on the accuracy of accelerometer-based estimates of vocal function. Methods: Simultaneous recordings were made of oral acoustic (microphone) and neck skin acceleration signals for 6 normal speakers and 18 patients with voice disorders (mild to severe dysphonia) as they performed several speech tasks. Measures of phonation time, fundamental frequency, and sound pressure level were extracted from the Two types of signals and compared. Results: It was generally demonstrated that accelerometer-based measures closely approximated corresponding measurements obtained from a microphone signal across all levels of dysphonia severity. Furthermore, there was evidence that in some cases the accelerometer may actually represent a more robust approach for estimating phonation parameters in disordered voices. Conclusions: The results generally support the recent application of accelerometers as phonation sensors in ambulatory voice monitoring systems that can be used in the clinical assessment and management of voice disorders.

Learning to Detect Vocal Hyperfunction From Ambulatory Neck-Surface Acceleration Features: Initial Results for Vocal Fold Nodules

Voice disorders are medical conditions that often result from vocal abuse/misuse which is referred to generically as vocal hyperfunction. Standard voice assessment approaches cannot accurately determine the actual nature, prevalence, and pathological impact of hyperfunctional vocal behaviors because such behaviors can vary greatly across the course of an individuals typical day and may not be clearly demonstrated during a brief clinical encounter. Thus, it would be clinically valuable to develop noninvasive ambulatory measures that can reliably differentiate vocal hyperfunction from normal patterns of vocal behavior. As an initial step toward this goal we used an accelerometer taped to the neck surface to provide a continuous, noninvasive acceleration signal designed to capture some aspects of vocal behavior related to vocal cord nodules, a common manifestation of vocal hyperfunction. We gathered data from 12 female adult patients diagnosed with vocal fold nodules and 12 control speakers matched for age and occupation. We derived features from weeklong neck-surface acceleration recordings by using distributions of sound pressure level and fundamental frequency over 5-min windows of the acceleration signal and normalized these features so that intersubject comparisons were meaningful. We then used supervised machine learning to show that the two groups exhibit distinct vocal behaviors that can be detected using the acceleration signal. We were able to correctly classify 22 of the 24 subjects, suggesting that in the future measures of the acceleration signal could be used to detect patients with the types of aberrant vocal behaviors that are associated with hyperfunctional voice disorders.

Talker-to-listener distance effects on speech production and perception

Simulating talker-to-listener distance (TLD) in virtual audio environments requires mimicking natural changes in vocal effort. Studies have identified several acoustic parameters manipulated by talkers when varying vocal effort. However, no systematic study has investigated vocal effort variations due to TLD, under natural conditions, and their perceptual consequences. This work examined the feasibility of varying the vocal effort cues for TLD in synthesized speech and real speech by (a) recording and analyzing single word tokens spoken at 1 m < or = TLD < or = 32 m, (b) creating synthetic and modified speech tokens that vary in one or more acoustic parameters associated with vocal effort, and (c) conducting perceptual tests on the reference, synthetic, and modified tokens to identify salient cues for TLD perception. Measured changes in fundamental frequency, intensity, and formant frequencies of the reference tokens across TLD were similar to other reports in the literature. Perceptual experiments that asked listeners to estimate TLD showed that TLD estimation is most accurate with real speech; however, large standard deviations in the responses suggest that reliable judgments can only be made for gross changes in TLD.

Smartphone-based detection of voice disorders by long-term monitoring of neck acceleration features

Many common voice disorders are chronic or recurring conditions that are likely to result from inefficient and/or abusive patterns of vocal behavior, termed vocal hyperfunction. Thus an ongoing goal in clinical voice assessment is the long-term monitoring of noninvasively derived measures to track hyperfunction. This paper reports on a smartphone-based voice health monitor that records the high-bandwidth accelerometer signal from the neck skin above the collarbone. Data collection is under way from patients with vocal hyperfunction and matched-control subjects to create a dataset designed to identify the best set of diagnostic measures for hyperfunctional patterns of vocal behavior. Vocal status is tracked from neck acceleration using previously-developed vocal dose measures and novel model-based features of glottal airflow estimates. Clinically, the treatment of hyperfunctional disorders would be greatly enhanced by the ability to unobtrusively monitor and quantify detrimental behaviors and, ultimately, to provide real-time biofeedback that could facilitate healthier voice use.

Temperature compensation of a quartz tuning-fork clock crystal via post-processing

The dependence of a tuning-fork quartz crystal oscillators frequency f on temperature T is observed over the temperature range -5 to 20°C. From this, a parabolic f(T) function is fit to the crystals data, and used to compensate for sampling period drift in an Analog to Digital Converter (ADC) system based around this crystal at various temperatures. Resolution and uncertainty of this method are discussed.

Accuracy of Self-Reported Estimates of Daily Voice Use in Adults With Normal and Disordered Voices

Purpose Accurate estimation of daily patterns of vocal behavior is essential to understanding the role of voice use in voice disorders. Given that clinicians currently rely on patient self-report to assess daily vocal behaviors, this study sought to assess the accuracy with which adults with and without voice disorders can estimate their amount of daily voice use in terms of phonation time. Method Eighteen subjects (6 patients, 6 matched members of a control group without voice disorders, 6 low voice users) wore the accelerometer-based Ambulatory Phonation Monitor (APM; model 3200, KayPENTAX, Montvale, NJ) for at least 5 workdays. Subjects were instructed to provide hourly self-reports of time spent talking using a visual analog scale. Spearman correlation coefficients and errors between self-reported and APM-based estimates of phonation time revealed subject- and group-specific characteristics. Results A majority of subjects exhibited a significant bias toward overestimating their phonation times, with an average absolute error of 113%. Correlation coefficients between self-reported and APM-based estimates of phonation time ranged from statistically nonsignificant to .91, reflecting large intersubject variability. Conclusions Subjects in all 3 groups were moderately accurate at estimating their hourly voice use, with a consistent bias toward overestimation. The results support the potential role that ambulatory monitoring could play in improving the clinical assessment of voice disorders.

Developing a portable and persistent autonomous real-time marine mammal acoustic monitor

Current methods for acoustically monitoring marine mammal habitats to mitigate against potential disruptions are compromised in their effectiveness due to non-real-time analysis, such as with archival recorders, or high system noise, such as with towed hydrophone arrays used with seismic surveys. To realize the advantages of both archival and real-time analysis systems, we are developing a portable and autonomous system for acquiring and analyzing towed hydrophone array data in real-time, by combining archival recording hardware, signal detection firmware, and high-bandwidth satellite communication onto a solar- and wave-powered glider platform. Such a system would be capable of persistent, autonomous, real-time monitoring of marine mammals in areas that would otherwise not be surveyed, as it will not require a local ship for its deployment, its retrieval, or reception of data for human review. This paper describes the ongoing development work to demonstrate the feasibility of integrating a WaveGlider with the archival recording electronics and a towed four-element hydrophone array to capture and output acoustic data to an on-ship data collection system.

Corrections to “Learning to Detect Vocal Hyperfunction From Ambulatory Neck-Surface Acceleration Features: Initial Results For Vocal Fold Nodules”

In, the third sentence of the second paragraph in Section III-D should have read as follows: “We first divided data using leave-one-out cross validation (LOOCV) to generate 12 subject subsets, where each subject subset consisted of randomly selected data across the 12 pairs. For each test subset, all windows from the 11 other subsets were then subdivided using fivefold cross validation (1/5th validation and 4/5th training in each fold).”

Improving acoustic time-of-arrival location estimates by correcting for temperature drift in time base oscillators

Using multiple acoustic sensors in an array for estimating sound source location relies on time synchrony among the devices. When independent time synchrony methods—such as GPS time stamps—are unavailable, the precision of the time base in individual sensors becomes one of the main sources of error in synchrony, and consequently increases the uncertainty of location estimates. Quartz crystal oscillators, on which many acoustic sensors base sampling rate timing, have a vibration frequency that varies with temperature f(T). Each oscillator exhibits a different frequency-temperature relationship, leading to sensor-dependent sample rate drift. Our Marine Autonomous Recording Units (MARUs) use such oscillators for their sample rate timing, and they experience variations in temperature of at least 20°C between preparation in air and deployment underwater, leading to sample rate drift over their deployments. By characterizing each MARU’s oscillator f(T) function, and measuring the temperature of the MARU during ...