Effect of local architectural acoustics on dyadic conversations over short distances

• Applicant: Manuj Yadav, Ph.D.
• Subject Area: Acoustics
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 503914237

Dyadic (two-person) conversations represent the simplest form of verbal communication. Within built spaces, it involves talking-listeners hearing their own and each other’s voices, room reflections, and background sounds. Compared to the Lombard effect wherein talking-listeners systematically raise their voice intensities (sound pressure levels (SPLs)) accompanied by spectral and prosodic changes with increasing background noise, studies of voice regulation due to room acoustics report subtler and inconsistent findings. The contributing factors include weak voice support (own voice reinforcement due to reflections) in typical rooms, and larger effects of background noise and the communication context. Yet, understanding basic factors underlying voice regulation during conversations is important. Recent findings indicate that strong voice support from local surfaces can reduce talkers’ voice intensities, hence reducing speech transmission to the surroundings. Retroreflective architectural surfaces, which reflect sound back to the source, have exhibited exciting prospects in providing such strong and local voice support, i.e., from nearby surfaces such as ceiling, wall treatments, etc., over short distances, compared to global voice support from room boundaries. However, the architectural design principles to effectively provide local voice support using retroreflectors are not fully understood. Nor do we know whether local voice support has beneficial effects during conversations.This project aims to (a) characterise voice regulation due to local voice support during dyadic conversations and, in turn, (b) optimise local voice support using retroreflecting and similar architectural surfaces. Aim (a) involves studying the linguistic-acoustic changes in interlocutor’s voices for short-distance dyadic communication scenarios for room acoustics and background noise variations, and for variations in voice support by combining physical (i.e., using fabricated retroreflectors) and/or virtual acoustics methods. Aim (b) includes studying physical acoustics of retroreflection using numerical simulations to characterize voice support as a function of architectural design aspects, room acoustics, background noise, and talker distances. Aim (b) informs the experimental design within (a) and is in turn informed by these findings to optimise voice support using retroreflection in an iterative manner. The project’s output will include predictive models of voice regulation due to factors in aims (a) and (b); and development of novel retroreflection and similar anomalous sound reflection phenomena within these factors. The expected speech SPL reduction due to local voice support – a positive behavioural effect of architectural acoustics – represents an innovative method of noise reduction in multi-talker areas including classrooms, offices, where traditional sound absorption methods are notoriously ineffective against speech distraction from nearby talkers.

• DFG Programme: Research Grants
• International Connection: Australia
• Co-Investigators: Professorin Dr.-Ing. Janina Fels; Professor Dr. Michael Vorländer
• Cooperation Partners: Dr. Joerg Buchholz; Dr. Densil Cabrera