Doctor of Philosophy
Health and Rehabilitation Sciences
Adams, Scott G.
Hypophonia (quiet speech) is a common speech symptom associated with Parkinson’s disease (PD), and is associated with reduced intelligibility, communicative effectiveness, and communicative participation. Studies of hypophonia commonly employ average speech intensity as the primary dependent measure, which may not entirely capture loudness deficits. Loudness may also be affected by the frequency components of speech (i.e. spectral balance) and speech level variability. The present investigation examined relationships between perceived loudness and intelligibility with acoustic measures of loudness, speech intensity, and spectral distribution in individuals with hypophonia secondary to Parkinson’s disease (IWPDs) and neurologically healthy older adults (HOAs).
Samples of sentence reading and conversational speech from 56 IWPDs and 46 HOAs were presented to listeners for ratings of perceived loudness and intelligibility. Listeners provided ratings of loudness using visual analogue scales (VAS) and direct magnitude estimation (DME). Acoustic measures of speech level (e.g. mean intensity), spectral balance (e.g. spectral tilt), and speech level variability (e.g. standard deviation of intensity) were obtained for comparison with perceived characteristics. In a spectral manipulation experiment, a gain adjustment altered the spectral balance of sentence samples while maintaining equal mean intensity. Listeners provided VAS ratings of perceived loudness of these manipulated samples.
IWPDs were quieter, less intelligible, and had a relatively greater concentration of low-frequency energy than HOAs. Speech samples with weaker contributions of mid- (2-5 kHz) and high-frequency (5-8 kHz) energy were perceived as quieter. Results of the spectral manipulation experiment indicated that increases in the relative contribution of 2-10 kHz energy were associated with increases in perceived loudness. The acoustic time-varying loudness model (TVL) demonstrated stronger associations with perceived loudness and larger differences between IWPDs and HOAs, and successfully identified differences in loudness in the spectral manipulation experiment. Loudness ratings provided with VAS and DME were consistent, both providing excellent reliability.
Findings of this investigation indicate that perceived loudness, acoustic loudness, and spectral balance are important components of hypophonia evaluation. Incorporating spectral manipulation in amplification by increasing high-frequency energy may improve efficacy of amplification devices for hypophonia management.
Summary for Lay Audience
Most individuals with Parkinson’s disease (PD) experience changes in their speech and voice. Quiet speech (hypophonia) is a common speech symptom associated with PD. Hypophonia interferes with the ability of individuals with PD (IWPDs) to effectively communicate because they may not be heard or understood, and some IWPDs may avoid communicating in situations they previously enjoyed. Effective assessment and evaluation of hypophonia is important in research and clinical settings to understand the condition and provide strategies to reduce the impacts of this condition on the lives of IWPDs. This study investigated several measures that can be used to assess hypophonia in order to identify components of effective assessment.
IWPDs and neurologically healthy older adults (HOAs) were recorded while reading sentences aloud and while participating in a conversation. Recordings of their speech were played for listeners, who rated the loudness and intelligibility (how much of their speech they could understand) of each sample. Acoustic measures were obtained from the speech recordings to compare how the sound characteristics of their speech related to the listeners’ perceptions. Acoustic algorithms designed to estimate perceived loudness were also included for comparison with perceived loudness. In a second experiment, listeners heard manipulated samples of speech. Frequency characteristics of speech were altered to investigate how the loudness would change.
IWPDs were quieter, less intelligible, and had disrupted spectral balance (frequency characteristics of their speech). Speech samples with relatively weaker high-frequency energy sounded quieter. Time-varying loudness (TVL; acoustic algorithm estimating loudness) provided effective measurement of loudness in both IWPDs and HOAs. Effective assessment of hypophonia may include listener judgments of loudness, acoustic calculations of loudness, and descriptions of spectral balance. Some IWPDs use amplification devices, similar to the microphone and loudspeaker used by a speaker in a large auditorium. Findings of this study suggest that incorporating a high-frequency boost to these amplifiers might further improve the loudness which would be a more effective tool for IWPDs with hypophonia.
Cushnie-Sparrow, Daryn Amory, "Modelling loudness: Acoustic and perceptual correlates in the context of hypophonia in Parkinson’s disease" (2021). Electronic Thesis and Dissertation Repository. 7862.