Towards a simultaneously speaking bilingual robot: Primary study on the effect of gender and pitch of the robot’s voice

Abstract

With fast and reliable international transportation, more people with different language backgrounds can interact now. As a result, the need for communicative agents fluent in several languages to assist those people is highlighted. The high cost of hiring human attendants fluent in several languages makes using social robots a more affordable alternative in international gatherings. A social robot capable of presenting a piece of information in more than one language at the same time to its audience is the goal of this line of study. However, the negative effect of background noise on speech comprehension in humans is well-established. Hence, presenting a piece of information in two different languages at the same time by the robot creates an adverse listening condition for both individuals listening to the speech of such a bilingual robot. In this study, we investigated whether manipulating the pitch and gender of the robot’s voice could affect human subjects’ memory of the presented information in the presence of background noise. The results indicate that the pitch and gender of the speaking voice do indeed affect our memory of the presented information. when a male voice was used, a higher pitch resulted in significantly better memory performance than a lower pitch. Contrarily, when a female voice was used, a lower pitch resulted in significantly better memory in participants than a higher pitch. Both male and female subjects performed significantly better with a female voice in a noisy background. In nutshell, the result of this study suggests using a female voice for robots in noisy conditions, as in the case of simultaneously speaking robots, can significantly improve the retrieval of presented information in human subjects.

Sound masking by a low-pitch speech-shaped noise improves a social robot’s talk in noisy environments

“…Hence, it is important to find ways to increase the efficiency of masking sounds by making them less annoying. In previous work, authors created a filter to create lowfrequency dominance in the Speech-Shaped Noise as an alternative to conventional Speech-Shaped Noise (Pourfannan et al, 2023). The logic behind the proposed filter was the body of evidence suggesting 10.3389/frobt.2023.1205209 that low-frequency ranges of the background noise contribute less to increasing the annoyance rate in the participants than the mid and high frequencies (Torija and Flindell, 2015).…”