Evaluating the effect of pauses on number recollection in synthesized speech

Abstract:

This study investigates the effects of an inserted pause on digit recollection for synthesized speech. Participants took part in a perception experiment which involved listening to a 7-digit random number that was rendered by a speech synthesis system. Some of the stimuli had pauses (200 ms or 500 ms in duration) inserted before one of the digits, while others did not include a pause. Immediately following each stimulus the participants were asked to provide a missing sequence of three adjacent digits. Results indicate that recall accuracy is improved immediately following a pause. Additionally, we found a significant effect for a pause duration of 500 ms but not for a pause duration of 200 ms. When investigating response time, we found that participants’ response time increased when a pause was present. Overall, the results show that pauses have a role to play in synthesized speech. This research can be regarded in the context of investigating pauses and pause-internal particles (e.g. breath noises) in synthesized speech and the effects they have for human listeners.


Year: 2021
In session: Sprachsynthese
Pages: 289 to 295