InterSpeech 2021

Analysis of Contextual Voice Changes in Remote Meetings
(3 minutes introduction)

Hector A. Cordourier Maruri (Intel, Mexico), Sinem Aslan (Intel, USA), Georg Stemmer (Intel, Germany), Nese Alyuz (Intel, USA), Lama Nachman (Intel, USA)
People participating in remote meetings in open spaces might choose to speak with a restrained voice due to concerns around privacy or disturbing others. These contextual voice changes might impact the quality of communications. To investigate how people adjust their voices in certain situations, we performed an exploratory data collection study with 41 participants in 18 simulated remote meetings. A scenario was provided to the participants to naturally trigger contextual voice changes. We collected multi-modal data from the participants including in-situ labels for the voice quality. We implemented content analysis, t-test, and linear regression to analyze the multi-modal data. Results showed that the participants primarily preferred to use soft voice over whispered voice to avoid being overheard during the meetings. Speaking softly was often sufficient to successfully conceal private conversations, while using whispered voice had only a negative impact on the intelligibility. Overall, we found that participants perceived soft voice as less pleasant to listen to than normal voice during meetings and discovered factors related to speaker demographics and meeting context that impacted the concealing behavior (soft or whispered). For our future research, we will expand to different scenarios and consider the impact of audio feedback on voice concealing.