13th Speech in Noise Workshop, 20-21 January 2022, Virtual Conference 13th Speech in Noise Workshop, 20-21 January 2022, Virtual Conference

P72 Who’s behind the mask? Facemasks interact with talker and listener characteristics to affect speech intelligibility in noise

Paulina Swiatek, Charles Masterson, Rachel Smith
University of Glasgow, UK

(a) Presenting
(b) Attending

Facemasks are a type of visual noise that can make speech harder to understand [Chládková et al., 2021, Psych. Bull. Rev 28(3):992; Toscano & Toscano, 2021, PLoS ONE 16(2):1]. We report two experiments to investigate how this visual noise affects speech intelligibility in combination with acoustic noise and with talker and listener characteristics: the listener’s autism-spectrum traits (Experiment 1), and talker familiarity (Experiment 2).

Experiment 1 investigated whether listeners with greater autism-spectrum traits are more affected by the presence of facemasks alongside noise (multitalker babble), compared to those with lesser autism-spectrum traits. 40 participants aged 19-25 completed an online survey consisting of an Autism Spectrum Quotient (AQ) questionnaire [Baron-Cohen, 2001, J. Autism Dev. Disord., 31:5] followed by video recordings of 20 sentences from the SPIN test [Kalikow, Stevens & Elliott 1977, JASA, 61:1337], which they attempted to transcribe. Presence/absence of a facemask and of noise were manipulated in a factorial within-participants design and counterbalanced across sentences. Sentence intelligibility was scored in terms of words correctly reported. At the group level, both noise and presence of a facemask reduced intelligibility. For each participant we separately calculated the drops in intelligibility they experienced due to presence of a facemask, and to noise. Participants’ AQ scores were significantly correlated with their intelligibility drop due to noise, and marginally correlated with the intelligibility drop due to a facemask. Experiment 2 investigated whether the listener’s personal familiarity with the talker affects intelligibility in cases where the talker is wearing a facemask. 16 participants each transcribed video recordings of 56 sentences from the SPIN test spoken by 4 talkers. Familiarity and presence of a facemask were manipulated in a within-participants design, and were counterbalanced across sentences. Each listener was personally familiar with two of the talkers and not with the other two; each talker appeared in both masked and unmasked conditions. Sentence intelligibility was again scored in terms of words correctly reported. While the unmasked conditions displayed high accuracy regardless of talker familiarity, in the masked conditions familiar talkers were more intelligible than unfamiliar talkers.

Taken together, the experiments show that the presence of a facemask interacts with and accentuates known adverse influences on speech intelligibility and allow us to identify situations where the wearing of facemasks is particularly likely to lead to comprehension difficulties.

Last modified 2022-01-24 16:11:02