Effects of Repeated Exposure to Synthetic Speech by Individuals with Aphasia

Abstract

For individuals with aphasia, comprehension challenges inhibit quality of life and daily task completion. Technological supports (e.g., text-to-speech (TTS)) offer methods to compensate for comprehension challenges by providing content through multiple modalities (i.e., visual and auditory inputs), which includes the use of computer-generated speech forms (synthetic speech). Although potentially beneficial, research suggests that individuals with aphasia comprehend synthetic speech output from TTS with varying degrees of success. At this time, it is unclear whether repeated exposure to synthetic speech increase comprehension of content presented through this modality. Therefore, the purpose of this project was to evaluate the effect of repeated exposure on comprehension of recorded natural speech (digitized speech) and various synthetic speech forms by people with aphasia. Participants included four adults, between 40 and 67 years of age, with a clinical aphasia diagnosis. Participants were exposed to one synthetic and one digitized speech output daily over a two-week period. Each day, participants listened to 30 simple, active sentences and provided a true/false response regarding the feasibility of each sentence. Participants then completed two follow-up sessions—one week and four weeks after experimental completion—to evaluate maintenance effects of repeated exposure and comprehension generalization to novel stimuli and untrained synthetic voice conditions. The results of this study revealed four main findings: (a) comprehension variability of the presented synthetic speech across and within individuals with aphasia, (b) increased comprehension of synthetic speech and digitized natural speech following repeated exposure to stimuli, (c) unique performance patterns dependent on aphasia profile, and (d) individualized and dynamic preference selection for synthetic speech. Findings from the study hold clinical implications when considering use of synthetic speech as a comprehension aid for individuals with aphasia. First of all, variable comprehension and preferences suggest trialing of multiple supports, and including unique patient preference when evaluating synthetic speech as a possible comprehension aid. Generally improved comprehension following repeated exposures, although not statistically significant, illustrates increased comprehension of synthetic and digitized natural speech over time among individuals with aphasia. Though improvements were not statistically significant, increases in comprehension may be clinically or personally significant for individuals using the support as a comprehension aid. Individuals with mild aphasia (n = 2) maintained high performance on comprehension tasks across time. However, performance of individuals with moderate aphasia (n = 2) fluctuated over time. This finding cautions against taking data at only a single timepoint, as it may not accurately reflect an individual’s auditory comprehension potential. Finally, the variability of performance on generalization tasks was highly individualized and suggested that generalization of auditory comprehension to an untrained task may be dependent on the individual. Collectively, these findings suggest that synthetic and digitized speech may be promising supports for individuals with mild and moderate aphasia. However, conclusions may not be generalizable to a wide range of individuals due to the limited participants in the current study. Further research is warranted to better understand the effects of repeated exposure on comprehension of synthetic and digitized speech by individuals with aphasia.