Acoustic profiles of speech amplification devices on hypophonic speech in Parkinson’s disease

Objective: Speech amplification devices are an augmentative management approach for hypophonia for some people with Parkinson’s disease (PD). There is a wide variety of commercially available amplification devices that speech-language pathologists report using with their clients with PD. However, the amplified speech signal’s acoustic profiles and resultant quality across devices may vary substantially, impacting potential treatment efficacy. The aim of the present study was to objectively characterize acoustic profiles of a range of commercially available amplification devices for management of hypophonia.

Methods: Seven commercially available speech amplification devices were calibrated and used to amplify three sets of audio stimuli: 1) pink noise, 2) sustained vowels, and 3) reading of a standardized passage. Speech recordings were obtained from an existing audio corpus from three male talkers with hypophonia secondary to PD. All audio stimuli were also amplified through a flat frequency response speaker, which served as a control. All amplified signals were audio recorded from a 1-meter distance. Spectral tilt and the maximum energy in three spectral frequency bands were measured from the long-term average spectrum of each audio recording: 0 – 1 kHz (low), 1 – 3 kHz (mid), and 3 – 8 kHz (high). Acoustic measures for each amplification device were compared to the baseline flat-frequency response to characterize the degree of spectral change.

Results: Compared to the baseline, all devices were associated with an overall decrease in low-frequency spectral energy and increases in mid- and high-frequency energy and spectral tilt.
While this overall pattern was consistent across stimuli, substantial variability in the magnitude of acoustic change was observed across the devices. The direction and magnitude of the acoustic change across devices were further influenced by baseline acoustic voice characteristics. For example, high-frequency energy in sustained vowels was attenuated, rather than amplified, by three of the seven devices for the talker with the lowest baseline.

Conclusions: Findings have clinical implications for speech amplification device selection and use for individuals with hypophonia secondary to PD. Results will be discussed in the context of auditory-perceptual implications and directions for technological improvements.