A preliminary laboratory assessment of simultaneous multilingual voice guidance based on the temporal and spatial locations of target speech

Abstract

In current multilingual public address systems, information is typically broadcast in multiple languages sequentially. This approach often results in significant delays, forcing users to wait for guidance in their native language. The present study investigates the optimization of simultaneous multilingual voice guidance to address this issue, focusing on the temporal and spatial locations of the target speech. A word intelligibility test, in which five consecutive words in up to four languages were presented simultaneously using single- or multi-loudspeaker systems, with Japanese as the target language and the other languages as interfering speech, showed that word intelligibility scores reached approximately 84% in the 4-stream condition and 90% in the 2- and 3-stream conditions. These results provide a basis for further studies aimed at the practical implementation of simultaneous multilingual voice guidance with up to four streams and identify key design considerations, including strategic placement of critical information in later temporal locations and spatial separation of the sources.