Abstract
Objectives:
1) Build into a commonly available mobile device the ability to accurately measure the acoustic characteristics of normal and abnormal voices. 2) Build a user interface that allows untrained patients to reproducibly measure the quality of their voice and track changes over time.
Methods:
Apple mobile devices (iPhones, iPads, iPod Touches) were chosen as the platform since they are equipped with microphones of consistent good quality for reliable acoustic voice analysis. Novel implementations of state of the art acoustic analysis algorithms best suited for these devices were developed. The algorithms and user interface were refined following usability tests on patients at University College London Hospitals from June 2012 to December 2012.
Results:
We have built into these devices the capability for performing analysis of (a) fundamental frequency (F0), formant frequencies (F1-F4), jitter and shimmer (b) maximum phonation time (c) reading and singing pitch ranges. Obtaining standardized recordings was hindered initially by excessive ambient noise. To minimize ambient noise, a volume level sensor that prevented the patient from completing an analysis outside of a quiet environment was implemented. Furthermore, visual and auditory feedback with spectrograms and playback was provided to the patient to check recording quality. Initial testing shows moderate to strong correlation and agreement to existing clinical voice quality measurement equipment.
Conclusions:
We have demonstrated that standardized acoustic analyses can be obtained on mobile devices without additional hardware or patient training. In this way, patients can be empowered to monitor their voice quality and engage with researchers and clinicians.
Get full access to this article
View all access options for this article.