Sydney Voice Lab engages in high quality, cutting-edge research pertaining to the voice and voice disorders. Our projects aim to enhance understanding of vocal physiology and its assessment through sensitive and specific voice measures.
We strive to improve the standards and efficiency of voice assessment, diagnosis, and specialist clinical care. Additionally, our goals are to establish the most effective means of treatment and to apply artificial intelligence in voice care, optimising patient outcomes.
Investigates the inner workings of the voice, including vocal physiology and the link between auditory perception and voice motor control. The research we conduct in this area is both experimental and clinical in nature. Inspired by Associate Professor Cate Madill's PhD 'The Semiotics of Voice Quality', we have continued exploring the impact of manipulating specific elements of the vocal system to assess vocal outcomes.
Our current projects in this theme involve measuring factors associated with voice production, the impact of disturbed voice production on self-perception of pitch and other vocal features, the effect of lung volume on phonation, and describing voice onset types and subsequent phonatory characteristics. By understanding the basic science of how the vocal system works, we can improve and evolve our assessment and treatment of voice disorders.
Voice disorders can be conceptualised into motor function of the vocal system and recognition of the perceptual outcomes of the voice. The interrelation between voice quality and pitch perception abilities is considered a new area of interest in the context of voice research.
This study seeks to identify whether the pitch perception and discrimination abilities of patients with voice disorders change as a result of undergoing voice therapy. The outcomes of this study will increase understanding of the relationship between the quality of a listener’s voice signal and their internal motor representations. Should a positive relationship between these factors be observed, this will provide evidence of a link existing between vocal perceptual and motor representations in the brain, as suggested by traditional perception and production models.
Past related studies
This study involves a comprehensive investigation into lung volume characteristics at the very onset of voicing, focusing on how trained and untrained speakers regulate lung volume across various vocal tasks. This project integrates detailed physiological data, including airflow, air pressure, electroglottography, and laryngeal resistance, alongside acoustic and perceptual voice measures.
The goal of this study is to develop a deeper understanding of phonatory control at different lung volume levels during key voicing phases: onset, steady phonation, and offset. This project has involved the development of innovative MATLAB scripts for precise data extraction. The dataset analysis is ongoing, with an emphasis on building a robust model that can inform both clinical and research applications. This work represents a significant step toward advancing both theoretical understanding and practical applications in voice science.
Past related studies
Chronic cough, being a cough which persists for an extended period of time (typically eight weeks or more), can have a debilitating impact upon the individual, their participation in all activities of daily living, and their quality of life.
This project focuses on the classification of chronic cough using cough threshold testing. Cough threshold testing essentially determines one’s cough sensitivity, by determining the lowest quantity of a particular irritant substance needed to trigger a cough in a given patient. The findings of this research will help to improve outcomes for individuals with chronic cough and related conditions.
Past related studies
This is a groundbreaking study aimed at exploring the nuanced voice and speech characteristics of gender-diverse Australians. This project aims to deepen our understanding of how gender-diverse individuals use their voices and speech patterns to express their gender identities and how these features influence listeners' perceptions of gender.
Unlike traditional studies that focus on binary gender distinctions, this research embraces the complexity of gender expression across a spectrum, offering a more inclusive perspective. The study integrates sociolinguistic and clinical research methodologies, examining the social, cultural, and physiological aspects of gendered speech, and uses machine learning techniques to analyse patterns in voice and speech data for a more data-driven understanding of how vocal characteristics relate to gender identity and expression. By shedding light on the complex connections between voice production, perception, speech, and gender, this study aims to contribute to more inclusive practices in clinical and educational environments.
Evaluating the various elements of the voice, using measures which are both sensitive and specific, is a crucial element of robust voice research. Our research in this domain seeks to improve the quality and efficiency of acoustic, aerodynamic, and perceptual vocal assessment measures, inclusive of fundamental frequency, voicing spectrograms, and voice onset, amongst others.
Acoustic analysis is an invaluable tool to validate and quantify auditory-perceptual voice quality, one of the key features of voice disorder. Acoustic analysis is also at the forefront of the revolution in artificial intelligence models for early detection of voice disorder and other diseases.
Led by Dr Duong Nguyen, we aim to explore the utility and scope of acoustic analysis in both research and clinical domains by testing the sensitivity and specificity of different acoustic measures in different contexts. This work will inform our use of acoustic analysis and assist clinicians and researchers alike to use the most appropriate acoustic measurements in the assessment and treatment of different voice disorders.
Past related studies
This project forms part of a broader branch of research within our program, exploring how perceptual judgements predict acoustic values. This specific work is a systematic review into the vocal tasks used for acoustic and/or auditory-perceptual analysis to discriminate individuals with and without voice disorders. Vocal tasks are defined activities or exercises which enable a voice clinician to gather information pertaining to a patient’s functional and/or physiological vocal limitations.
The multidimensionality of voice has led to the development of a great diversity of vocal tasks currently in existence, both within and between vocal assessment categories, with varying degrees of evidence supporting their value and utility in measuring different aspects of voice quality. With the current plethora of vocal tasks reported across the literature, this review seeks to provide the initial step in elucidating the range of acoustic and auditory-perceptual vocal tasks present across the evidence base, and the value of each in discriminating voice disorders from normophonic voicing.
Past related studies
Earlier research conducted by our research team evaluated global practice patterns for the initial diagnostic assessment of voice disorders. This revealed an alignment with published guidelines but with notable variability in the selection and application of assessment tools. Case history was rated as the most important assessment element and decisions regarding which other assessments were used were based on case history information. Despite its importance, no guidelines for case history taking/questionnaires exist in the current literature.
These findings underscore the need to explore the role of case history in voice disorder diagnosis and served as the springboard for the conception of our current study, aimed at developing a standardised case history questionnaire for the collection of consistent, essential diagnostic patient information across varied voice clinical settings.
Past related studies
This is a prospective research project aimed at examining how phonation onset characteristics influence subsequent steady-state phonation, in collaboration with researchers from Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany. This study involves enrolling vocally trained and untrained participants and gathering a comprehensive dataset that spans acoustic, aerodynamic, physiological, and high-speed visualisation data across different phonation onsets.
The goal of this project is to develop a novel voice onset measurement tool for clinical use and establish effective vocal manipulation techniques to address vocal dysfunction. Thus far, this research has led to the development of a new voice onset measurement tool, to be made publicly available following the project’s completion. Participant recruitment is ongoing, and full data collection will commence shortly using our new high-speed imaging system. This research represents a significant step forward in advancing both clinical practices and therapeutic interventions in voice research.
Past related studies
This project represents a collaboration with researchers from Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany, to develop, test, and validate the Voice Onset Analysis Tool (VOAT), a new voice onset measurement tool designed for both research and clinical applications.
The VOAT program was published in SoftwareX in 2024. Currently, VOAT is being validated using larger datasets from individuals with and without voice disorders. The tool is integral to two of our major ongoing projects: measuring voice onset at different lung volumes and evaluating the impact of various window lengths and algorithms on voice onset. Data collected using VOAT for these projects is expected to be published in 2025. Further development of this project will focus on automating the extraction of voice onset data for clinical use, with this phase set for implementation in 2026.
Past related studies
This project aims to develop a tool which captures acoustic vowel characteristics in real-time. These characteristics are known as 'formants' and they provide information about a speaker’s vowel articulation. This software tool will record pre-set words and provide immediate feedback, evaluating whether the speaker’s vowel production matches normative data for their gender and accent background.
This software program and the findings of this project are expected to benefit the field of speech-language pathology by making accurate acoustic vowel assessment fast and accessible for clinicians and enabling the use of objective acoustic measures in clinical settings. This work is conducted in collaboration with Professor Michael Döllinger from Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany, and employs a master's program student, Masudur Rahman Kazi.
Past related studies
In healthcare, professional bodies are groups of professionals from the same or related disciplines and expertise. Professional bodies play an essential role in advancing clinical practices and establishing field-related guidelines pertaining to practice recommendations, scope of practice, and standard of care. This study involves conducting a website review of professional bodies related to the field of voice and voice disorders to determine what recommendations are available for ENT specialists and speech-language pathologists.
The findings of this review aim to inform professional bodies of the need to advocate for and support practice recommendations for voice clinicians. This will allow for these recommendations to be made more widely accessible and also support the development of voice assessment clinical practice guidelines with a higher evidence base.
The rate at which artificial intelligence technologies and applications are being utilised in healthcare is rising exponentially. Working alongside The University of Sydney’s Faculty of Engineering, we have made considerable progress in investigating the potential of artificial intelligence for the early detection and diagnosis of voice disorders. Our research seeks to harness and maximise the potential of this technology to improve all elements of the voice patient’s management pathway, from diagnosis through to treatment.
Across both speech-language pathology and otorhinolaryngology, an extensive range of terms and approaches are used in diagnosing a voice disorder. There is an inherent lack of standardisation which can generate confusion and a lack of clarity across both clinical and research settings. This in turn can impact the effective management of a patient with vocal concerns.
This study seeks to use artificial intelligence and machine learning to explore innovative algorithmic approaches to enhance diagnostic accuracy. A critical element of this research will involve developing a standardised system of voice disorder classification, which has the potential to transform patient voice assessment and diagnosis on a global scale.
This project relates to a comprehensive scoping review of current artificial intelligence algorithms utilised in voice disorder detection. This has entailed a thorough examination of scholarly articles on the application of machine learning in voice disorder recognition; a process that has laid a solid foundation for further, more expansive studies in the artificial intelligence voice classification domain.
A notable achievement in this project’s research journey has been the acceptance of our scoping review protocol by BMJ Open, underscoring the growing recognition and impact of our work in advancing the field of voice disorder diagnostics. This review has thus far identified considerable variations in voice disorder classification diagnostic labels, data availability, and methodology. Such variation limits comparability and may be a critical barrier to clinical application, a finding which could pave the way for a revived approach to voice disorder artificial intelligence systems in future research.
Past related studies
The establishment of a large clinical voice database with a comprehensive range of data types and reliable diagnostic labelling would create a significant opportunity to expand clinical voice research outcomes and provide much needed data for the development of artificial models for early detection of voice disorders. It would help set the standard for voice clinical care on a global scale and provide the big data required to generate research outcomes of increased statistical power and higher confidence in our results. The University of Sydney Voice Lab seeks to develop such a database and is working on a number of projects to bring this databank to fruition.
Our current active research studies, which are laying important groundwork in the establishment of this database, include a project evaluating the quality of current voice databases, a survey of speech-language pathologists' and ear, nose, and throat specialists' perspectives pertaining to voice databases, and an exploration of the perceptions and willingness of the public towards the establishment of an online digital voice biobank.
Assessment of the voice is a crucial element of the patient management pathway. In spite of this, one of the greatest barriers some patients with voice disorders face is timely, appropriate access to care from voice specialists.
Our research into facilitated access to specialist voice assessment and care seeks to improve the quality and efficiency of voice patient management pathways and voice assessment methods to ensure that all patients with voice disorders can benefit from the most efficacious management pathway for their presenting vocal concerns.
This research relates to an ever-evolving body of work pertaining to patient access to voice specialist clinical care. It commenced with the completion of a PhD study exploring the speech-language pathology primary contact (SLPPC) model of voice assessment. In this alternative pathway, people with vocal symptoms who are referred to an ear, nose, and throat (ENT) surgeon are seen by a speech-language pathologist prior to case discussion with an ENT for complete diagnosis. This pathway was found to improve wait times for people in the public health system, initiate multidimensional voice assessment at an earlier stage, and identify people suitable for ENT management at a faster rate.
This body of research has now evolved into an exploration of how the components of a multidimensional virtual voice assessment can predict voice disorder classification and the urgency for ENT examination. These projects are helping to reimagine voice patient clinical pathways in a way which improves access and in turn, benefits patients’ clinical outcomes.
Past related studies
Both the ear, nose, and throat specialist (ENT) and the speech-language pathologist play a critical role in improving the function and participation of patients with voice disorders. In the field of ENT, our research delves into new avenues in otolaryngological treatments and surgical approaches for voice disorder patients, making a significant impact upon their voice-related quality of life.
Speech pathology therapy entails behavioural change and exercises which transform the way in which a patient uses their voice. Our research in this domain seeks to explore the most effective means to deliver targeted and effective speech pathology voice treatment, enabling those with voice disorders to engage fully and functionally in all elements of daily life.
Understanding how we learn to manipulate and change the vocal system is key to improving behavioural treatments. The Motor Learning Classification Framework (MLCF), developed by Madill et. al., describes a schema of actions and processes used in teaching motor skills in voice and other areas of speech-language pathology. This model continues to be explored and expanded in a number of projects, including assessing the impact of negative practice versus repetitive drills, and applying the MLCF to classical singing. Research outcomes will empower speech-language pathologists and voice teachers to reflect on and improve the efficacy of their practice, thereby improving outcomes for patients and students alike.
Past related studies
Clinical investigations have been conducted to evaluate the efficacy and safety of various ear, nose, and throat treatment regimens for patients with chronic cough and laryngeal sensory dysfunction. These treatments include neuromodulators, superior laryngeal nerve block, and laryngeal botulinum toxin injections.
This study examines the treatment effects of neuromodulators in managing chronic refractory cough. In the case of chronic cough, which occurs as a result of an overactive cough reflex, neuromodulators help to reduce cough severity by improving regulation of nerve signals. The outcomes of this study will help to advance therapeutic strategies and improve outcomes for individuals with chronic cough and related conditions.
Past related studies
This project aims to investigate the technique of THRIVE for oxygenation during laryngeal surgery.
This will involve measuring the real-time laryngeal oxygen concentration using THRIVE at 100% oxygen concentration followed by either 30% oxygen concentration or apneic conditions (i.e., without breathing). The oxygen saturation/desaturation rates and vital characteristics using THRIVE at different levels of oxygen concentration delivery and in apneic conditions will be examined and compared.
The findings of this project will help elucidate the most appropriate processes for patient oxygenation during laryngeal surgery and could help transform practices on an international scale.
Past related studies
Sob Voice Therapy is a direct voice therapy program used to treat muscle tension voice disorder and is adapted in gender affirming voice therapy for voice feminisation. Despite being used in clinics over the last 20 years, the efficacy of this treatment has not been thoroughly investigated.
This research program aims to investigate the biomechanical and perceptual elements that underpin the stages of the program and test the outcomes of the therapy in a series of clinical trials, beginning with retrospective analysis of clinical data and moving towards randomised control trials in both populations.
Past related studies
The University of Sydney Voice Lab team is always open to new and diverse ways of achieving our program’s mission, including involvement in local and international projects which depart from our core research theme areas but represent an opportunity to enhance and expand our knowledge in voice-related areas.
A significant cornerstone of our research program is collaboration, and as such, we have continued to embrace a range of varied projects to further advance knowledge in the field of voice disorders and laryngology. Research projects which fall under this theme reflect our openness to collaboration as one of the world’s leading voice research laboratories.
Speech-language pathology practice is mediated by the interaction between a client and their clinician. To develop an effective therapeutic alliance and provide therapy which addresses the client’s communication needs, it is essential that the speech pathology practitioner has highly developed clinical communication skills. However, communication skills are rooted in culture, and as such, cultural considerations are of utmost importance to consider when training clinicians from diverse linguistic and cultural backgrounds.
This project examines and compares perceptual ratings of voice quality and clinical communication behaviours between two cohorts of speech pathology students in Sydney and Hong Kong. The findings of this study will clarify whether differences in culture and language background affect perception of voice and clinical communication, which may facilitate specific teaching and learning strategies to help Chinese students work effectively with English speaking clients, and Australian students to work with Chinese speaking clients.
Gender-nonconforming voice and speech features, such as higher and more varied fundamental frequency as well as /s/ and vowel production with higher frequencies, have been examined as markers of gay male identities in multiple varieties of English. Yet production of these features is not observed consistently and limited results are available for Australian English (AusE).
To expand our understanding of gay male voice and speech to AusE, we investigated gay-straight differences in voice and speech production by measuring fundamental frequency, /s/, and vowel production in AusE using data from 35 gay and 28 straight speakers. We further explored speech differences within gay men based on their self-perceived masculinity and variables relevant to their sexual identities (e.g., outness, internalised homophobia). Our findings have the potential to support voice clinical outcomes in the LGBTQI+ community and will help pave the way for further research of this demographic in an Australian context.
Past related studies
This project explores a proposed framework for audio voice forensics, which includes both manual and automated state-of-the-art artificial intelligence analysis techniques to facilitate speaker profiling and identification of sounds. With the exponential increase in access to mobile smart devices, use of cloud-based platforms, and digital surveillance, the quantity of audio voice criminal evidence continues to rise. This presents us with a unique opportunity to make a meaningful impact in the world of forensic investigation through leveraging our team’s experience and cross-discipline in audio voice analysis.
At present, there is a lack of transparency pertaining to methodological approaches to voice forensic analysis. This study presents a voice forensics investigation framework to enable a robust cross-examination of acoustic data without succumbing to the pitfalls of confirmation bias. Led by Dr Brian Stasak, this project has great potential to significantly impact the scientific weight of audio voice forensics outcomes in the courtroom and to strengthen confidence in our conclusions.
This study will evaluate how well people estimate individuals' ages and gender when listening to voice recordings. The conceptualisation of this project arose from a small-scale study requested by the Australian Office of the Commonwealth Director of Public Prosecutions (CWPP) last year, which played a critical role in a forensic case. Audio recording case evidence can involve short duration unintelligible utterances (e.g., laughs, sighs, filler sounds) in non-adult verbal populations. Yet, vital information concerning real world non-adult related audio voice forensics is extremely scarce in the present literature. The outcomes of this investigation will provide influential voice forensic insights regarding perceptual age-gender estimation capabilities based on an underrepresented demographic.
Below are our top 10 most highly cited papers published during the program:
Email
voice.lab@sydney.edu.au
Mailing address
Level 5
Susan Wakil Health Building
University of Sydney
Camperdown, NSW 2006
