Electronic Thesis and Dissertation Repository

Thesis Format

Monograph

Degree

Master of Science

Program

Surgery

Supervisor

Fung, Kevin

2nd Supervisor

Doyle, Philip

Co-Supervisor

3rd Supervisor

Parsa, Vijay

Co-Supervisor

Abstract

Background:

Acoustic voice analysis requires a resource intensive setup, including a soundproof booth. This project evaluates smartphone microphone and recording environment impacts on voice sample collection for acoustic voice analysis. A proprietary analysis algorithm is presented for validation.

Methods:

Microphone and recording environment were evaluated using previously collected voice samples presented in four conditions to test two microphones and two recording environments. Prospective samples were used to test the proprietary algorithm, whereby samples were analyzed using this and Praat.

Results:

Microphone and recording environment had small, clinically unimportant impacts on most measurements. The proprietary algorithm reliably analyzed sustained vowels, with strong correlation to the Praat results. Continuous speech analysis was less reliable.

Conclusion:

Smartphone microphones are adequate for voice sample collection. Quiet, non-soundproof settings can be used for voice collection. The proprietary algorithm represents a reliable method to analyze sustained vowel samples. Some improvements are necessary before continuous speech analysis can be considered valid.

Summary for Lay Audience

Acoustic voice analysis is a part of a thorough voice examination, which can be used in conjunction with aerodynamic measurements, auditory-perceptual analysis, and patient reported outcomes. It is a method of objectively quantifying normal and pathologic voices. It relies on physical properties of voice and the resultant vocal signal.

The existing methods of voice sample collection and analysis are often resource intensive, creating barriers to access. In the current work, we sought to empirically examine a multi-phase study evaluating smartphones as an avenue to perform voice sample acquisition and on-device acoustic voice analysis.

In phase one, microphones for voice sample collection and the influence of recording environment were evaluated. This showed small, clinically unimportant differences in most acoustic measures for both the microphone and the recording environment. These results indicate that current barriers to voice signal acquisition can be relaxed upon, without degrading the ability to measure microacoustic changes. In conclusion, the results of this phase indicate that robust acoustic voice analysis is feasible on current day smartphones with samples collected under non-soundproof conditions.

In phase two, a proprietary application, developed for the purposes of this study, was compared against the current standard used for acoustic voice analysis (Praat). The results of this study provide data, based on a systematic acquisition methods, which supports future use of the current proprietary algorithm. In doing so, the current findings indicate that the present smartphone application could be used reliably by patients, healthcare providers, and researchers for acoustic voice analysis.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Download

Share

COinS